The latest articles on DEV Community by ironbyte-rgb (@ironbyte-rgb). https://dev.to/ironbyte-rgb https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3968914%2Fc3f6aee1-e034-4e41-9f5b-bfbe1937e61b.png https://dev.to/ironbyte-rgb en ironbyte-rgb Tue, 14 Jul 2026 20:00:15 +0000 https://dev.to/crescevo/developers-ditch-github-over-tech-issues-and-politics-m8f https://dev.to/crescevo/developers-ditch-github-over-tech-issues-and-politics-m8f <h2> TL;DR </h2> <ul> <li>GitHub hosts over 600 million repositories and had nearly one billion commits in 2025.</li> <li>Some high-profile projects, such as Ghostty and Zig, have abandoned GitHub due to concerns over technical problems and the service's political direction.</li> <li>Developers are exploring alternatives, including Codeberg and self-hosting options.</li> <li>Projects like Tenacity have already made the switch, now only maintaining a mirror presence on GitHub.</li> </ul> <p>GitHub remains a popular platform, with one new user joining every second, but some users are concerned about issues like frequent downtime and the service's direction since being taken over by Microsoft. According to Bobby Jack, a technology enthusiast, some key players have abandoned GitHub, potentially marking the beginning of a wider exodus. The Linux kernel source code, for example, has a read-only GitHub mirror, but its main home is hosted on its own domain.</p> <h2> What the data shows </h2> <p>The data indicates that while GitHub is still widely used, some developers are seeking alternatives due to various concerns. Mitchell Hashimoto, the maintainer of Ghostty, announced in April 2026 that the project would be leaving GitHub, citing the need to remove dependencies on the platform. Zig, a system programming language, also departed from GitHub in November 2025, after an uninterrupted run since 2015. These moves suggest that some developers are dissatisfied with GitHub's current state.</p> <h2> What this means for tech readers </h2> <p>For tech readers, this trend may signal a shift towards more decentralized and self-hosted solutions. Codeberg, a GitHub alternative, is one option being explored by developers. Tenacity, a cross-platform audio editor, has already made the switch, now only maintaining a mirror presence on GitHub. This could be an indication that more projects will follow suit, potentially leading to a more diverse landscape of development platforms.</p> <h2> What to do right now </h2> <p>Developers looking for alternatives to GitHub can explore options like Codeberg, which offers a similar set of features and tools. Self-hosting is also a viable option, allowing developers to maintain full control over their projects and data. For those interested in exploring these alternatives, it's essential to research and evaluate the pros and cons of each option, considering factors like ease of use, scalability, and community support.</p> <h2> Bottom line </h2> <p>In summary, while GitHub remains a widely used platform, some developers are seeking alternatives due to concerns over technical problems and the service's direction. The departure of high-profile projects like Ghostty and Zig may mark the beginning of a wider trend towards decentralized and self-hosted solutions. As the development landscape continues to evolve, it's essential for developers to stay informed about the options available and make informed decisions about their projects and data.</p> <h2> Frequently asked questions </h2> <h3> Q: Why are some developers leaving GitHub? </h3> <p>Some developers are leaving GitHub due to concerns over technical problems, such as frequent downtime, and the service's direction since being taken over by Microsoft.</p> <h3> Q: What alternatives are available to GitHub? </h3> <p>Alternatives to GitHub include Codeberg, a GitHub alternative, and self-hosting options, which allow developers to maintain full control over their projects and data.</p> <h3> Q: Which high-profile projects have abandoned GitHub? </h3> <p>High-profile projects that have abandoned GitHub include Ghostty, a cross-platform terminal emulator, and Zig, a system programming language.</p> <h3> Q: What does this trend mean for the future of development platforms? </h3> <p>This trend may signal a shift towards more decentralized and self-hosted solutions, potentially leading to a more diverse landscape of development platforms.</p> <h2> Sources </h2> <ul> <li><a href="https://www.howtogeek.com/why-developers-are-ditching-github-for-codeberg-and-self-hosting-alternatives/?ref=tech.crescevo.com" rel="noopener noreferrer">https://www.howtogeek.com/why-developers-are-ditching-github-for-codeberg-and-self-hosting-alternatives/</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://tech.crescevo.com/chatto-is-now-open-source/" rel="noopener noreferrer">Chatto opens source for self-hosted encrypted team chat</a></li> <li><a href="https://tech.crescevo.com/the-arm-takeover-is-complete-what-it-means-for-developers/" rel="noopener noreferrer">The ARM Takeover Is Complete — What It Means for Developers</a></li> <li><a href="https://tech.crescevo.com/typescript-7/" rel="noopener noreferrer">TypeScript 7 runs 10x faster natively</a></li> </ul> <p><em>Originally published on <a href="https://tech.crescevo.com/why-developers-are-ditching-github-for-c/" rel="noopener noreferrer">Tech at Crescevo</a> — subscribe free for more.</em></p> programming devops webdev tech ironbyte-rgb Tue, 14 Jul 2026 19:00:15 +0000 https://dev.to/crescevo/claude-code-wastes-26k-more-tokens-than-opencode-per-prompt-5f4i https://dev.to/crescevo/claude-code-wastes-26k-more-tokens-than-opencode-per-prompt-5f4i <h2> TL;DR </h2> <ul> <li>Claude Code sends 33,000 tokens before reading the prompt, while OpenCode sends 7,000 tokens.</li> <li>Claude Code is far more cache inefficient than OpenCode, rewriting tens of thousands of prompt-cache tokens mid-session.</li> <li>A small task that cost 121,000 tokens done directly cost 513,000 tokens when fanned out to two subagents.</li> <li>Claude Code's whole-task total came out lower than OpenCode's on a multi-step task, but this advantage did not hold on a newer model.</li> </ul> <p>Claude Code and OpenCode were tested on the same model, machine, and tasks to examine their token usage. The results show that Claude Code sends significantly more tokens before reading the prompt than OpenCode. This has significant implications for the efficiency and cost of using these systems.</p> <h2> What the data shows </h2> <p>The data shows that Claude Code uses roughly 33,000 tokens of system prompt, tool schemas, and injected scaffolding before the prompt even arrives, while OpenCode uses about 7,000. Additionally, Claude Code re-writes tens of thousands of prompt-cache tokens mid-session, run after run, and on the same task writes up to 54x more cache tokens than OpenCode. This cache inefficiency results in higher costs, as cache writes are billed at a premium.</p> <h2> What this means for ai readers </h2> <p>The high token usage of Claude Code has significant implications for AI readers. For example, a production repository's 72KB instruction file adds another 20,000 tokens to every single request. Five modest MCP servers add 5,000 to 7,000 more tokens. By the time a real working setup sends its first request, it is 75,000 to 85,000 tokens deep before the user has typed a word. This can result in higher costs and reduced efficiency.</p> <h2> What to do right now </h2> <p>To minimize token usage and reduce costs, it is essential to optimize the system prompt and tool schemas. This can be achieved by caching the payload once per session and reading it back for pennies, as OpenCode does. Additionally, using a more efficient system like OpenCode can help reduce token usage and costs. It is also crucial to consider the impact of subagents on token usage, as fanning out tasks to multiple subagents can multiply the number of full baselines in flight.</p> <h2> Bottom line </h2> <p>The results of the test show that Claude Code sends significantly more tokens before reading the prompt than OpenCode, resulting in higher costs and reduced efficiency. While Claude Code's whole-task total came out lower than OpenCode's on a multi-step task, this advantage did not hold on a newer model. To minimize token usage and reduce costs, it is essential to optimize the system prompt and tool schemas and consider using a more efficient system like OpenCode.</p> <h2> Frequently asked questions </h2> <h3> Q: How many tokens does Claude Code send before reading the prompt? </h3> <p>Claude Code sends roughly 33,000 tokens before reading the prompt.</p> <h3> Q: How does OpenCode's token usage compare to Claude Code's? </h3> <p>OpenCode sends about 7,000 tokens before reading the prompt, significantly less than Claude Code's 33,000 tokens.</p> <h3> Q: What is the impact of subagents on token usage? </h3> <p>Fanning out tasks to multiple subagents can multiply the number of full baselines in flight, resulting in higher token usage and costs. For example, a small task that cost 121,000 tokens done directly cost 513,000 tokens when fanned out to two subagents.</p> <h3> Q: Can Claude Code's whole-task total be lower than OpenCode's on certain tasks? </h3> <p>Yes, Claude Code's whole-task total came out lower than OpenCode's on a multi-step task, but this advantage did not hold on a newer model.</p> <h2> Sources </h2> <ul> <li><a href="https://systima.ai/blog/claude-code-vs-opencode-token-overhead?ref=ai.crescevo.com" rel="noopener noreferrer">https://systima.ai/blog/claude-code-vs-opencode-token-overhead</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://ai.crescevo.com/gpt-5-6-sol-terra-luna-government-preview/" rel="noopener noreferrer">GPT-5.6 Went to the Government Before It Goes to You</a></li> <li><a href="https://ai.crescevo.com/claude-corps/" rel="noopener noreferrer">Claude Corps</a></li> <li><a href="https://ai.crescevo.com/making-claude-a-chemist/" rel="noopener noreferrer">Making Claude a Chemist</a></li> </ul> <p><em>Originally published on <a href="https://ai.crescevo.com/claude-code-sends-33k-tokens-before-read/" rel="noopener noreferrer">AI at Crescevo</a> — subscribe free for more.</em></p> ai machinelearning llm programming ironbyte-rgb Mon, 13 Jul 2026 20:00:15 +0000 https://dev.to/crescevo/typescript-7-runs-10x-faster-natively-3ha6 https://dev.to/crescevo/typescript-7-runs-10x-faster-natively-3ha6 <h2> TL;DR </h2> <ul> <li>TypeScript 7 is a 10x faster native port of TypeScript.</li> <li>The new version brings native code speed, shared memory multithreading, and optimizations that yield speedups between 8x and 12x on full builds.</li> <li>TypeScript 7 is available via npm and supports the language server protocol (LSP) for editor support.</li> <li>The new version can be installed using the command npm install -D typescript.</li> </ul> <p>TypeScript 7 has been announced, bringing significant performance improvements to the JavaScript development process. The new version is a native port of TypeScript built in Go, which makes the most of modern hardware. According to the announcement, TypeScript 7 brings a 10x faster performance compared to its predecessor.</p> <h2> What the data shows </h2> <p>The data shows that TypeScript 7 is a significant improvement over previous versions, with speedups between 8x and 12x on full builds. The new version also brings native code speed and shared memory multithreading, which enables faster development and editing. The announcement highlights that the new codebase was written while maintaining the structure and logic of the original codebase to keep results consistent and compatible between the two compilers.</p> <h2> What this means for tech readers </h2> <p>For tech readers, TypeScript 7 means a faster and more efficient development process. A typical day of development will be streamlined, with features like auto-completions and red squiggles on the fly working faster than before. The new version also supports the language server protocol (LSP), which enables editor support for a wide range of code editors, including VS Code, Visual Studio, and WebStorm.</p> <h2> What to do right now </h2> <p>To get started with TypeScript 7, users can install it using the command npm install -D typescript. This will get them the new tsc executable in their workspace, which can be run via npx tsc. Users can also check their editor's documentation to see if it supports TypeScript 7, as many modern editors have dedicated extensions or automatic support for the new version.</p> <h2> Bottom line </h2> <p>In summary, TypeScript 7 is a significant improvement over previous versions, bringing a 10x faster performance and native code speed. The new version supports the language server protocol (LSP) and can be installed using npm. With its faster development process and wide range of editor support, TypeScript 7 is a must-have for any JavaScript developer.</p> <h2> Frequently asked questions </h2> <h3> Q: How can I install TypeScript 7? </h3> <p>TypeScript 7 can be installed using the command npm install -D typescript.</p> <h3> Q: What are the performance improvements in TypeScript 7? </h3> <p>TypeScript 7 brings speedups between 8x and 12x on full builds, as well as native code speed and shared memory multithreading.</p> <h3> Q: Does TypeScript 7 support my code editor? </h3> <p>TypeScript 7 supports the language server protocol (LSP), which enables support for a wide range of code editors, including VS Code, Visual Studio, and WebStorm.</p> <h3> Q: Is TypeScript 7 compatible with previous versions? </h3> <p>Yes, TypeScript 7 is compatible with previous versions, as the new codebase was written while maintaining the structure and logic of the original codebase.</p> <h2> Sources </h2> <ul> <li><a href="https://devblogs.microsoft.com/typescript/announcing-typescript-7-0/?ref=tech.crescevo.com" rel="noopener noreferrer">https://devblogs.microsoft.com/typescript/announcing-typescript-7-0/</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://tech.crescevo.com/rewriting-bun-in-rust/" rel="noopener noreferrer">Bun ditches Zig for Rust to boost performance</a></li> <li><a href="https://tech.crescevo.com/chatto-is-now-open-source/" rel="noopener noreferrer">Chatto opens source for self-hosted encrypted team chat</a></li> <li><a href="https://tech.crescevo.com/my-thoughts-on-the-bun-rust-rewrite/" rel="noopener noreferrer">Bun's VC funding shifted creator Jarred from open source to high stress</a></li> </ul> <p><em>Originally published on <a href="https://tech.crescevo.com/typescript-7/" rel="noopener noreferrer">Tech at Crescevo</a> — subscribe free for more.</em></p> programming devops webdev tech ironbyte-rgb Mon, 13 Jul 2026 19:00:15 +0000 https://dev.to/crescevo/alec-scollon-dreads-llm-output-despite-boosted-productivity-die https://dev.to/crescevo/alec-scollon-dreads-llm-output-despite-boosted-productivity-die <h2> TL;DR </h2> <ul> <li>Alec Scollon's usage rate of LLMs is probably average by current dev standards.</li> <li>He spends hours each day interacting with LLMs, including Claude Code and Codex.</li> <li>Scollon feels more productive with LLMs but has started to dread reading LLM output due to false assumptions and hallucinations.</li> <li>He thinks continually learning how to use LLMs effectively is valuable, despite feeling LLM burnout.</li> </ul> <p>Alec Scollon's job has changed significantly with the integration of LLMs, from designing and writing code to designing code, describing the design to an LLM, reviewing code, and writing code. He uses LLMs like Claude Code and Codex for tasks, and also relies on them for casual queries, often using ChatGPT or reading Gemini's overview. However, he has started to feel a sense of dread when reading LLM output due to the repetition of false assumptions and hallucinations.</p> <h2> What the data shows </h2> <p>Alec Scollon's experience with LLMs has been extensive, with hours of interaction each day. He has worked with various LLMs, including Claude Code, Codex, and Qwen, and has used them for tasks such as creating tooling and sifting through unsupervised agent output. The data shows that Scollon's usage rate of LLMs is probably average by current dev standards, and his methods are probably primitive. He has also started to feel more comfortable in areas where he doesn't have deep knowledge, thanks to the LLM steps that expose him to new approaches.</p> <h2> What this means for ai readers </h2> <p>For AI readers, Scollon's experience highlights the potential for LLM burnout. The repetition of false assumptions and hallucinations in LLM output can be wearing, even for those who find LLMs valuable. Scollon's use of personalization features to mitigate this issue is notable, but the fact that some idiosyncrasies still seep through suggests that more work is needed to improve the user experience. AI readers may need to develop strategies to deal with the limitations of LLMs and find ways to use them effectively without feeling overwhelmed.</p> <h2> What to do right now </h2> <p>To address LLM burnout, users like Scollon may need to take a step back and reassess their usage of LLMs. This could involve taking breaks from LLM output, seeking out diverse sources of information, and developing strategies to mitigate the repetition of false assumptions and hallucinations. Additionally, LLM developers may need to prioritize improving the quality and diversity of LLM output to reduce the risk of burnout. Scollon's experience suggests that continually learning how to use LLMs effectively is valuable, but it is also important to be aware of the potential for burnout and take steps to prevent it.</p> <h2> Bottom line </h2> <p>Alec Scollon's experience with LLM burnout highlights the need for users and developers to be aware of the potential risks and limitations of LLMs. While LLMs can be valuable tools, the repetition of false assumptions and hallucinations can be wearing. By prioritizing improvement and developing strategies to mitigate these issues, users and developers can work together to create a better experience for everyone.</p> <h2> Frequently asked questions </h2> <h3> Q: What is LLM burnout? </h3> <p>LLM burnout refers to the feeling of exhaustion or frustration that can come from repeatedly reading LLM output that contains false assumptions and hallucinations.</p> <h3> Q: How does Alec Scollon use LLMs? </h3> <p>Alec Scollon uses LLMs like Claude Code and Codex for tasks, and also relies on them for casual queries, often using ChatGPT or reading Gemini's overview.</p> <h3> Q: What are some potential solutions to LLM burnout? </h3> <p>Potential solutions to LLM burnout include taking breaks from LLM output, seeking out diverse sources of information, and developing strategies to mitigate the repetition of false assumptions and hallucinations.</p> <h3> Q: Why is it important to address LLM burnout? </h3> <p>Addressing LLM burnout is important because it can help users and developers work together to create a better experience for everyone, and prioritize improvement and development of LLMs to reduce the risk of burnout.</p> <h2> Sources </h2> <ul> <li><a href="https://www.alecscollon.com/blog/llm-burnout/?ref=ai.crescevo.com" rel="noopener noreferrer">https://www.alecscollon.com/blog/llm-burnout/</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://ai.crescevo.com/i-think-anthropic-and-openai-have-found/" rel="noopener noreferrer">I think Anthropic and OpenAI have found product-market fit</a></li> <li><a href="https://ai.crescevo.com/the-mysterious-hy3-llm-is-topping-openro/" rel="noopener noreferrer">Hy3 LLM tops OpenRouter rankings despite inferior model quality</a></li> <li><a href="https://ai.crescevo.com/real-time-llm-inference-on-standard-gpus/" rel="noopener noreferrer">Real-time LLM Inference on Standard GPUs: 3k tokens/s per request</a></li> </ul> <p><em>Originally published on <a href="https://ai.crescevo.com/i-think-i-have-llm-burnout/" rel="noopener noreferrer">AI at Crescevo</a> — subscribe free for more.</em></p> ai machinelearning llm programming ironbyte-rgb Mon, 13 Jul 2026 15:00:15 +0000 https://dev.to/crescevo/top-remote-tech-jobs-this-week-july-13-2026-3608 https://dev.to/crescevo/top-remote-tech-jobs-this-week-july-13-2026-3608 <p>The best remote Tech jobs this week. All roles are fully remote. Updated July 13, 2026.</p> <h3> <strong>Grow the Roses — Interface Designer</strong> </h3> <p><em>c · music · game dev · dev · video</em></p> <p>About The RoleWe are looking for a talented Interface Designer to create intuitive, visually engaging, and user-friendly digital experiences. You will work closely with UX designers, developers, and p...</p> <p><a href="https://remoteok.com/remote-jobs/remote-interface-designer-grow-the-roses-1134742?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Joko — CRM &amp; Lifecycle Manager</strong> </h3> <p><em>exec · design · sys admin · teaching · technical</em></p> <p>At Joko, we help consumers shop smarter. Our mission is to revolutionize shopping, empowering people to find what they need, make informed decisions, and save money.Founded in Paris, Joko is a tech co...</p> <p><a href="https://remoteok.com/remote-jobs/remote-crm-lifecycle-manager-joko-1134710?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Digital Style House — Social Media Coordinator</strong> </h3> <p><em>c · music · game dev · dev · video</em></p> <p>Company Description Digital Style House (DSH) is an Inc. 5000-ranked social media marketing agency based in Los Angeles, serving over 100 active clients and 1,100+ brands. Founded in 2016, DSH operate...</p> <p><a href="https://remoteok.com/remote-jobs/remote-social-media-coordinator-digital-style-house-1134743?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>INNERGY — Director of Customer Success</strong> </h3> <p><em>c · music · game dev · dev · video</em></p> <p>INNERGY is transforming the woodworking industry with purpose built, cloud based ERP software designed exclusively for custom manufacturers. Our platform helps businesses operate smarter, faster, and...</p> <p><a href="https://remoteok.com/remote-jobs/remote-director-of-customer-success-innergy-1134738?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>DataBank — Billing Specialist</strong> </h3> <p><em>exec · strategy · customer support · testing · content writing</em></p> <p>DataBank Holdings Ltd. is a leading provider of enterprise-class data center, cloud, and interconnection services, offering customers 100% uptime availability of data, applications, and infrastructure...</p> <p><a href="https://remoteok.com/remote-jobs/remote-billing-specialist-databank-1134749?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Lingraphica — Educational Programs Administrator</strong> </h3> <p><em>c · music · game dev · dev · video</em></p> <p>Lingraphica is a mission-driven organization that provides speech-generating devices to help improve communication, speech, and quality of life for people with communication impairments. Lingraphica i...</p> <p><a href="https://remoteok.com/remote-jobs/remote-educational-programs-administrator-lingraphica-1134741?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Crevtus — Designer</strong> </h3> <p><em>c · music · game dev · dev · video</em></p> <p>Location: Remote FullReports To: Lead Brand StrategistWhat We NeedA Designer that will be responsible for creating still designs for the brands we work with in the form of social creatives and larger...</p> <p><a href="https://remoteok.com/remote-jobs/remote-designer-crevtus-1134739?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Conquer AI — Application</strong> </h3> <p><em>infosec · dev · photoshop · digital nomad · education</em></p> <p>Don't See Your Role? Apply Anyway!Conquer AIRemote FirstFull-time Across all time zones, if you can make it work then apply.We are building out clusters globally.Must speak fluent EnglishAbout the job...</p> <p><a href="https://remoteok.com/remote-jobs/remote-application-conquer-ai-1134673?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>KAST — KYC Compliance Analyst</strong> </h3> <p><em>c · music · game dev · dev · video</em></p> <p>Application Deadline: 17 July 2026Department: Legal &amp; ComplianceLocation: TampaDescriptionWe are looking for a A KYC (Know Your Customer) Analyst who is a compliance professional responsible for v...</p> <p><a href="https://remoteok.com/remote-jobs/remote-kyc-compliance-analyst-kast-1134740?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>VENDED — We’re not hiring right now — but we’re growing fast</strong> </h3> <p><em>ios · dev · web dev · api · sales</em></p> <p>RoleIf you’d like us to consider you for future roles at VENDED, send us your CV and a few details. We’ll review every submission and keep strong matches on file:Engineering, Hardware &amp; Firmwa...</p> <p><a href="https://remoteok.com/remote-jobs/remote-were-not-hiring-right-now-but-were-growing-fast-vended-1134690?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>INNERGY — Customer Success Manager</strong> </h3> <p><em>design · hr · sys admin · infosec · teaching</em></p> <p>INNERGY is transforming the woodworking industry with purpose built, cloud based ERP software designed exclusively for custom manufacturers. Our platform helps businesses operate smarter, faster, and...</p> <p><a href="https://remoteok.com/remote-jobs/remote-customer-success-manager-innergy-1134651?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Moda Health — Health Navigator I</strong> </h3> <p><em>design · hr · sys admin · infosec · teaching</em></p> <p>Let’s do great things, together!About ModaFounded in Oregon in 1955, Moda is proud to be a company of real people committed to quality. Today, like then, we’re focused on building a better future...</p> <p><a href="https://remoteok.com/remote-jobs/remote-health-navigator-i-moda-health-1134652?ref=tech.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <p>Subscribe to the Tech Brief for weekly job picks and industry signals.</p> <p><em>Originally published on <a href="https://tech.crescevo.com/remote-tech-jobs-2026-07-13/" rel="noopener noreferrer">Tech at Crescevo</a> — subscribe free for more.</em></p> programming devops webdev tech ironbyte-rgb Mon, 13 Jul 2026 14:00:15 +0000 https://dev.to/crescevo/top-remote-ai-ml-jobs-this-week-july-13-2026-2k4p https://dev.to/crescevo/top-remote-ai-ml-jobs-this-week-july-13-2026-2k4p <p>The best remote AI &amp; ML jobs this week. All roles are fully remote. Updated July 13, 2026.</p> <h3> <strong>ReLytics Hire — Data Entry Specialist Assistant Administrator</strong> </h3> <p><em>web3 · crypto · blockchain · analyst · infosec</em></p> <p>Remote Data Entry Specialist - Assistant Administrator Work Mode: Remote We are seeking a Remote Data Entry Specialist - Assistant Administrator to support data management and administrative operation...</p> <p><a href="https://remoteok.com/remote-jobs/remote-data-entry-specialist-assistant-administrator-relytics-hire-1134602?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>ReLytics Hire — Customer Support</strong> </h3> <p><em>web3 · crypto · blockchain · analyst · infosec</em></p> <p>Customer Support Specialist (Arabic) | Remote  We are actively seeking a detail-oriented Remote Customer Support (Arabic) Specialist to join our team. In this role, you will be responsible for deliv...</p> <p><a href="https://remoteok.com/remote-jobs/remote-customer-support-relytics-hire-1134600?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Crypto.com — Compliance Market Surveillance Analyst</strong> </h3> <p><em>web3 · crypto · blockchain · analyst · infosec</em></p> <p>Our Local Programs and Regulatory Compliance team is responsible for ensuring compliance with local laws and regulations in the countries where we operate.Crypto.com | Derivatives North America (“CD...</p> <p><a href="https://remoteok.com/remote-jobs/remote-compliance-market-surveillance-analyst-crypto-com-1134599?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>RecruitGo Careers — CAD Assistant</strong> </h3> <p><em>web3 · crypto · blockchain · analyst · infosec</em></p> <p>About Us:RecruitGo is a leading Employer of Record (EOR) company specializing in bridging top talents in emerging markets with global clients. Our services enable businesses to hire, manage, and pay e...</p> <p><a href="https://remoteok.com/remote-jobs/remote-cad-assistant-recruitgo-careers-1134601?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Noventra Group — Paid Media Specialist Meta &amp; Google Ads</strong> </h3> <p><em>Marketing · Paid Media · Meta Ads · Facebook Ads · Google Ads</em></p> <p>Noventra Group is hiring a remote Paid Media Specialist to support Meta Ads and Google Ads campaigns across selected partner-brand initiatives.This role is focused on performance marketing, campaign o...</p> <p><a href="https://remoteok.com/remote-jobs/remote-paid-media-specialist-meta-google-ads-noventra-group-1134552?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Lemon.io — Senior AI Engineer Architect</strong> </h3> <p><em>dev · engineer · senior · full time · ai</em></p> <p>Are you a talented Senior AI Engineer/Architect looking for a remote job that lets you show your skills and get decent compensation? Look no further than Lemon.io — the marketplace that connects you...</p> <p><a href="https://remoteok.com/remote-jobs/remote-senior-ai-engineer-architect-lemon-io-1134396?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>ATOM PARTNERS LIMITED — Junior Crypto Trader</strong> </h3> <p><em>other · crypto · finance · web3 · bitcoin</em></p> <p>ATOM Partners is an international company focused on digital asset markets, trading technologies, and data-driven market research. We work with modern analytical tools and help aspiring professionals...</p> <p><a href="https://remoteok.com/remote-jobs/remote-junior-crypto-trader-atom-partners-limited-1134351?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>EverAI — Mid Senior AI Cinematic Video Editor</strong> </h3> <p><em>design · ai · other · art direction</em></p> <p>Our Vision &amp; Products🚀 EverAI — Building the Future of AI CompanionshipOne of the Top 15 Largest &amp; Fastest-Growing AI Companies in the World50 Million Users in 2 years — Help Us Reach 1...</p> <p><a href="https://remoteok.com/remote-jobs/remote-mid-senior-ai-cinematic-video-editor-everai-1134014?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>NOPE — Head of Public Understanding</strong> </h3> <p><em>marketing · communications · content writing · ai · part time</em></p> <p>Please note: this is a part-time contracting role.People are increasingly turning to AI for real relationships. They use it for companionship, for coaching, for emotional support, for advice, and some...</p> <p><a href="https://remoteok.com/remote-jobs/remote-head-of-public-understanding-nope-1133914?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>RECSEEKERS — Operations Administrator</strong> </h3> <p><em>hr · virtual assistant · medical · customer support · amazon</em></p> <p>EDTECH OPERATIONS ADMINISTRATOR<br> FULLY REMOTE | $75,000 - $110,000 BASE SALARY + BONUS +…See this and similar jobs on LinkedIn.Please mention the word <strong>RETRACTABLE</strong> and tag RNS43OC4xODguMjE= when a...</p> <p><a href="https://remoteok.com/remote-jobs/remote-operations-administrator-recseekers-1133675?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Swoon — Graphic Designer</strong> </h3> <p><em>hr · virtual assistant · medical · customer support · amazon</em></p> <p>Location: Fully RemoteDuration: 6-Month Contract with Potential ExtensionSchedule: Monday–Friday | Standard Business HoursAbout the OpportunityThis role is ideal for a highly creative presentation d...</p> <p><a href="https://remoteok.com/remote-jobs/remote-graphic-designer-swoon-1133676?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <h3> <strong>Blackstone Talent Group — Scrum Master</strong> </h3> <p><em>hr · virtual assistant · medical · customer support · amazon</em></p> <p>Scrum Master (Contractor) — Job DescriptionPosition Details:Location: Dallas, Tx (Remote)Type: 6 Month Contract - NO C2CResponsibilities: Support Agile delivery for assigned team(s) under guidance o...</p> <p><a href="https://remoteok.com/remote-jobs/remote-scrum-master-blackstone-talent-group-1133679?ref=ai.crescevo.com" rel="noopener noreferrer"><strong>Apply →</strong></a></p> <p>Subscribe to the AI &amp; ML Brief for weekly job picks and industry signals.</p> <p><em>Originally published on <a href="https://ai.crescevo.com/remote-ai-jobs-2026-07-13/" rel="noopener noreferrer">AI at Crescevo</a> — subscribe free for more.</em></p> ai machinelearning llm programming ironbyte-rgb Sun, 12 Jul 2026 20:00:15 +0000 https://dev.to/crescevo/bun-ditches-zig-for-rust-to-boost-performance-4bn4 https://dev.to/crescevo/bun-ditches-zig-for-rust-to-boost-performance-4bn4 <h2> TL;DR </h2> <ul> <li>Bun was initially written in Zig and has gained significant traction with over 22 million monthly downloads.</li> <li>The project's scope has also presented challenges for stability, with numerous bugs fixed in recent versions.</li> <li>Rewriting Bun in Rust is a significant undertaking, with the potential to improve performance and reliability.</li> <li>Bun's acquisition by Anthropic in December 2025 may influence the project's future development and direction.</li> </ul> <p>Bun's journey began as a line-for-line port of esbuild's JavaScript and TypeScript transpiler from Go to Zig, with the first line of Zig written on April 16, 2021. The project's massive scope includes a JavaScript, TypeScript, and CSS transpiler, minifier, and bundler, as well as an npm-compatible package manager and Jest-like test runner. The initial version of Bun was written by a single person in just one year, a feat made possible by the use of Zig.</p> <h2> What the data shows </h2> <p>The data shows that Bun has experienced significant growth, with its CLI receiving over 22 million monthly downloads. Popular tools like Claude Code and OpenCode have also bet on Bun as their runtime, and companies like Vercel, Railway, and DigitalOcean have implemented first-party support for Bun. However, the project's scope has also presented challenges, with numerous bugs fixed in recent versions, including heap-use-after-free crashes, use-after-free crashes, and memory leaks.</p> <h2> What this means for tech readers </h2> <p>For tech readers, the rewriting of Bun in Rust is a significant development that may improve the project's performance and reliability. The use of Rust could provide better memory safety and concurrency support, addressing some of the challenges faced by the current Zig implementation. Additionally, the acquisition of Bun by Anthropic in December 2025 may bring new resources and expertise to the project, potentially accelerating its development and adoption.</p> <h2> What to do right now </h2> <p>For those interested in Bun, now is a good time to explore the project's current capabilities and limitations. The primary source article provides a detailed overview of the project's history, scope, and challenges, as well as the motivations behind the decision to rewrite it in Rust. Readers can also visit the Bun website to learn more about the project's features and use cases, and to stay up-to-date with the latest developments.</p> <h2> Bottom line </h2> <p>In summary, the rewriting of Bun in Rust is a significant undertaking that may improve the project's performance and reliability. With its massive scope and significant traction, Bun has the potential to become a major player in the tech industry. As the project continues to evolve, it will be interesting to see how the use of Rust and the acquisition by Anthropic influence its development and adoption.</p> <h2> Frequently asked questions </h2> <h3> Q: What is Bun and what does it do? </h3> <p>Bun is a JavaScript, TypeScript, and CSS transpiler, minifier, and bundler, as well as an npm-compatible package manager and Jest-like test runner.</p> <h3> Q: Why was Bun initially written in Zig? </h3> <p>Bun was initially written in Zig because of its low-level control and care for performance, which allowed the project's creator to build a significant amount of functionality in a short amount of time.</p> <h3> Q: What are the challenges faced by the current Zig implementation of Bun? </h3> <p>The current Zig implementation of Bun has faced numerous challenges, including heap-use-after-free crashes, use-after-free crashes, and memory leaks, which have been addressed in recent versions.</p> <h3> Q: How may the acquisition of Bun by Anthropic influence the project's development and direction? </h3> <p>The acquisition of Bun by Anthropic may bring new resources and expertise to the project, potentially accelerating its development and adoption, and influencing its future direction and priorities.</p> <h2> Sources </h2> <ul> <li><a href="https://bun.com/blog/bun-in-rust?ref=tech.crescevo.com" rel="noopener noreferrer">https://bun.com/blog/bun-in-rust</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://tech.crescevo.com/my-thoughts-on-the-bun-rust-rewrite/" rel="noopener noreferrer">Bun's VC funding shifted creator Jarred from open source to high stress</a></li> <li><a href="https://tech.crescevo.com/chatto-is-now-open-source/" rel="noopener noreferrer">Chatto opens source for self-hosted encrypted team chat</a></li> <li><a href="https://tech.crescevo.com/ibm-sub-1nm-nanostack-chip/" rel="noopener noreferrer">IBM Sub-1nm Chip: Ignore the Number, the Real Story Is Going Vertical (and It Is 5 Years Out)</a></li> </ul> <p><em>Originally published on <a href="https://tech.crescevo.com/rewriting-bun-in-rust/" rel="noopener noreferrer">Tech at Crescevo</a> — subscribe free for more.</em></p> programming devops webdev tech ironbyte-rgb Sun, 12 Jul 2026 19:00:15 +0000 https://dev.to/crescevo/mistrals-8b-robostral-navigate-outperforms-multi-sensor-robots-41h5 https://dev.to/crescevo/mistrals-8b-robostral-navigate-outperforms-multi-sensor-robots-41h5 <h2> TL;DR </h2> <ul> <li>Mistral's Robostral Navigate is an 8B model that enables robots to autonomously navigate complex environments using a single RGB camera.</li> <li>The model achieves 76.6% success on unseen R2R-CE benchmarks, outperforming multi-sensor approaches.</li> <li>Robostral Navigate generalizes across robot types and adapts to real-world obstacles unseen during training.</li> <li>The model combines pointing-based navigation with reinforcement learning for continuous improvement.</li> </ul> <p>Mistral's Robostral Navigate is a state-of-the-art robotics navigation model that allows robots to navigate complex environments using only a single RGB camera. The model has achieved a success rate of 76.6% on unseen R2R-CE benchmarks, outperforming other approaches that use multiple sensors. This technology has numerous applications across manufacturing, delivery, logistics, and hospitality.</p> <h2> What the data shows </h2> <p>The data shows that Robostral Navigate has achieved state-of-the-art performance on R2R-CE, with a success rate of 79.4% on validation seen and 76.6% on validation unseen. The model operates from a single RGB camera, with no LiDAR or depth sensors, and has been trained entirely in simulation. The model's performance is robust to differences in camera intrinsics and can run on wheeled, legged, and flying robots, generalizing across robot sizes.</p> <h2> What this means for ai readers </h2> <p>For AI readers, Robostral Navigate's performance means that robots can now navigate complex environments with greater ease and accuracy. The model's ability to generalize across robot types and adapt to real-world obstacles unseen during training makes it a significant advancement in the field of robotics. The use of a single RGB camera also makes the model more efficient and cost-effective compared to other approaches that require multiple sensors.</p> <h2> What to do right now </h2> <p>To take advantage of Robostral Navigate's capabilities, users can provide the model with a plain-language instruction, such as "Leave the lobby, walk through the corridor, enter the supply room, and stop to face the second shelf." The model will then navigate the robot through the environment to complete the task. Users can also customize the model to work with different types of robots and environments, making it a versatile solution for a variety of applications.</p> <h2> Bottom line </h2> <p>The bottom line is that Mistral's Robostral Navigate is a state-of-the-art robotics navigation model that has achieved impressive results in navigating complex environments using a single RGB camera. The model's performance, efficiency, and versatility make it an attractive solution for a wide range of applications, from manufacturing and delivery to logistics and hospitality.</p> <h2> Frequently asked questions </h2> <h3> Q: What type of sensor does Robostral Navigate use? </h3> <p>Robostral Navigate uses a single RGB camera, with no LiDAR or depth sensors.</p> <h3> Q: What is the success rate of Robostral Navigate on unseen R2R-CE benchmarks? </h3> <p>The success rate of Robostral Navigate on unseen R2R-CE benchmarks is 76.6%.</p> <h3> Q: Can Robostral Navigate work with different types of robots? </h3> <p>Yes, Robostral Navigate can run on wheeled, legged, and flying robots, and generalizes across robot sizes.</p> <h3> Q: How does Robostral Navigate adapt to real-world obstacles unseen during training? </h3> <p>Robostral Navigate combines pointing-based navigation with reinforcement learning for continuous improvement, allowing it to adapt to real-world obstacles unseen during training.</p> <h2> Sources </h2> <ul> <li><a href="https://mistral.ai/news/robostral-navigate/?ref=ai.crescevo.com" rel="noopener noreferrer">https://mistral.ai/news/robostral-navigate/</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://ai.crescevo.com/unified-controllable-and-faithful-text-t/" rel="noopener noreferrer">LLMs now control CAD generation with state-of-the-art faithfulness</a></li> <li><a href="https://ai.crescevo.com/the-mysterious-hy3-llm-is-topping-openro/" rel="noopener noreferrer">Hy3 LLM tops OpenRouter rankings despite inferior model quality</a></li> <li><a href="https://ai.crescevo.com/claude-fable-5-mythos-5-launch/" rel="noopener noreferrer">Anthropic Launches Claude Fable 5 and Mythos 5: A New Model Class Above Opus</a></li> </ul> <p><em>Originally published on <a href="https://ai.crescevo.com/mistral-s-robostral-navigate-a-state-of/" rel="noopener noreferrer">AI at Crescevo</a> — subscribe free for more.</em></p> ai machinelearning llm programming ironbyte-rgb Sat, 11 Jul 2026 19:00:15 +0000 https://dev.to/crescevo/chatto-opens-source-for-self-hosted-encrypted-team-chat-bpn https://dev.to/crescevo/chatto-opens-source-for-self-hosted-encrypted-team-chat-bpn <h2> TL;DR </h2> <ul> <li>Chatto, a group and team chat application, is now open source and available for self-hosting.</li> <li>Chatto can be easily installed through Homebrew with the command "brew install chattocorp/tap/chatto".</li> <li>Chatto prioritizes data protection and privacy, with all personal and chat data fully encrypted at rest with per-user keys.</li> <li>Chatto Cloud, a paid hosting service for Chatto servers, will soon enter public beta.</li> </ul> <p>Chatto's transition to open source is a significant development, as it allows users to self-host the application on their own infrastructure. According to the announcement, Chatto is designed to be extremely easy to self-host, with a simple executable that serves its own frontend. The application prioritizes data protection and privacy, making it an attractive option for users who value security.</p> <h2> What the data shows </h2> <p>The primary source article highlights several key features of Chatto, including its compact and snappy design, as well as its support for voice and video calls with screen-sharing. The application is fully end-to-end encrypted, and calls can scale to as many participants as the infrastructure can handle. Additionally, Chatto's self-hosting capabilities allow users to host multiple communities by simply spinning up multiple Chatto processes.</p> <h2> What this means for tech readers </h2> <p>For tech-savvy readers, Chatto's open source status and self-hosting capabilities make it an appealing option for group and team chat needs. The application's emphasis on data protection and privacy is also a significant advantage, particularly in an era where data security is a top concern. Furthermore, Chatto's support for voice and video calls with screen-sharing makes it a versatile tool for remote communication and collaboration.</p> <h2> What to do right now </h2> <p>To get started with Chatto, users can install it through Homebrew using the command "brew install chattocorp/tap/chatto", followed by "chatto init" and "chatto run". The Chatto Getting Started Guide provides more detailed instructions for self-hosting the application. Alternatively, users can wait for the public beta launch of Chatto Cloud, which will offer paid hosting for Chatto servers.</p> <h2> Bottom line </h2> <p>In summary, Chatto's transition to open source is a significant development that offers users a secure, versatile, and easy-to-use group and team chat application. With its emphasis on data protection and privacy, as well as its support for voice and video calls with screen-sharing, Chatto is an attractive option for users who value security and collaboration.</p> <h2> Frequently asked questions </h2> <h3> Q: What is Chatto and what does it do? </h3> <p>Chatto is a group and team chat application that allows users to communicate and collaborate with each other. It offers features such as voice and video calls with screen-sharing, and prioritizes data protection and privacy.</p> <h3> Q: How can I install Chatto? </h3> <p>Chatto can be installed through Homebrew using the command "brew install chattocorp/tap/chatto", followed by "chatto init" and "chatto run". The Chatto Getting Started Guide provides more detailed instructions for self-hosting the application.</p> <h3> Q: What is Chatto Cloud and what does it offer? </h3> <p>Chatto Cloud is a paid hosting service for Chatto servers that will soon enter public beta. It offers simple, ad-free hosting for Chatto servers, with no premium subscriptions or third-party tracking or analytics.</p> <h3> Q: Is Chatto secure and private? </h3> <p>Yes, Chatto prioritizes data protection and privacy, with all personal and chat data fully encrypted at rest with per-user keys. The application is also fully end-to-end encrypted, making it a secure option for users who value security and privacy.</p> <h2> Sources </h2> <ul> <li><a href="https://www.hmans.dev/blog/chatto-is-open-source?ref=tech.crescevo.com" rel="noopener noreferrer">https://www.hmans.dev/blog/chatto-is-open-source</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://tech.crescevo.com/my-thoughts-on-the-bun-rust-rewrite/" rel="noopener noreferrer">Bun's VC funding shifted creator Jarred from open source to high stress</a></li> <li><a href="https://tech.crescevo.com/microsoft-s-6502-basic-is-now-open-sourc/" rel="noopener noreferrer">Microsoft's 6502 BASIC is now Open Source (2025)</a></li> <li><a href="https://tech.crescevo.com/ibm-sub-1nm-nanostack-chip/" rel="noopener noreferrer">IBM Sub-1nm Chip: Ignore the Number, the Real Story Is Going Vertical (and It Is 5 Years Out)</a></li> </ul> <p><em>Originally published on <a href="https://tech.crescevo.com/chatto-is-now-open-source/" rel="noopener noreferrer">Tech at Crescevo</a> — subscribe free for more.</em></p> programming devops webdev tech ironbyte-rgb Fri, 10 Jul 2026 20:00:15 +0000 https://dev.to/crescevo/buns-vc-funding-shifted-creator-jarred-from-open-source-to-high-stress-10g7 https://dev.to/crescevo/buns-vc-funding-shifted-creator-jarred-from-open-source-to-high-stress-10g7 <h2> TL;DR </h2> <ul> <li>Jarred, the creator of Bun, has strong "beginner energy" which helped him learn and grow in the Zig community.</li> <li>After taking venture capital, Jarred's focus shifted from working on a free and open source project to running a business.</li> <li>The "beginner energy" that was once seen as a positive trait started to have a different impact when Jarred became a manager.</li> <li>Jarred's management style, including demanding a poor work-life balance, has been a topic of discussion among those who interviewed for or worked at Oven.</li> </ul> <p>Andrew Kelley's thoughts on the Bun Rust rewrite highlight the shift in Jarred's approach after taking venture capital. Jarred's initial appreciation for the Zig project, including a monthly donation of $60,000 per year to the Zig Software Foundation, is notable. However, as Jarred became a manager, his "beginner energy" started to have a different impact on those around him.</p> <h2> What the data shows </h2> <p>The primary source article highlights Jarred's background and his involvement in the Zig community. Jarred joined the Zig community about 5 years ago and was described as having strong "beginner energy". This energy helped him learn and grow, but also led to mediocre outcomes in terms of engineering. As Jarred focused on Bun, he began to attract attention, and his project became a promising new toolchain. Jarred's decision to take venture capital instead of crowdfunding had a significant impact on his approach.</p> <h2> What this means for tech readers </h2> <p>The shift in Jarred's approach after taking venture capital is a key takeaway from the article. As Jarred became a manager, his focus shifted from working on a free and open source project to running a business. This change had a significant impact on those around him, including those who interviewed for or worked at Oven. The article highlights the importance of considering the impact of venture capital on the approach and values of a project or company.</p> <h2> What to do right now </h2> <p>The article does not provide specific recommendations for what to do right now. However, it highlights the importance of considering the potential impact of venture capital on a project or company. Tech readers may want to consider the potential consequences of taking venture capital and the importance of maintaining a focus on the values and approach that are core to their project or company.</p> <h2> Bottom line </h2> <p>The article provides a nuanced view of Jarred's approach and the impact of taking venture capital on his project. The shift in Jarred's approach after taking venture capital is a key takeaway, and tech readers should consider the potential consequences of this type of funding. The article highlights the importance of considering the values and approach that are core to a project or company and the potential impact of external factors on these values.</p> <h2> Frequently asked questions </h2> <h3> Q: What is "beginner energy"? </h3> <p>"Beginner energy" refers to the attitude of someone who is new to a field or project and is eager to learn and try new things. This energy can be beneficial for learning and growth, but can also lead to mediocre outcomes in terms of engineering.</p> <h3> Q: Why did Jarred take venture capital instead of crowdfunding? </h3> <p>The article does not provide a specific reason why Jarred took venture capital instead of crowdfunding. However, it notes that Jarred was groomed in the Silicon Valley mindset and may have been influenced by this approach.</p> <h3> Q: What is the impact of Jarred's management style on those around him? </h3> <p>Jarred's management style, including demanding a poor work-life balance, has been a topic of discussion among those who interviewed for or worked at Oven. The article suggests that this approach has had a negative impact on those around him.</p> <h3> Q: What can tech readers learn from Jarred's experience? </h3> <p>Tech readers can learn the importance of considering the potential impact of venture capital on a project or company. The article highlights the need to consider the values and approach that are core to a project or company and the potential consequences of external factors on these values.</p> <h2> Sources </h2> <ul> <li><a href="https://andrewkelley.me/post/my-thoughts-bun-rust-rewrite.html?ref=tech.crescevo.com" rel="noopener noreferrer">https://andrewkelley.me/post/my-thoughts-bun-rust-rewrite.html</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://tech.crescevo.com/ibm-sub-1nm-nanostack-chip/" rel="noopener noreferrer">IBM Sub-1nm Chip: Ignore the Number, the Real Story Is Going Vertical (and It Is 5 Years Out)</a></li> <li><a href="https://tech.crescevo.com/apple-price-hike-ai-memory-shortage/" rel="noopener noreferrer">Apple Just Raised Prices Mid-Cycle. The AI Boom Finally Reached Your Laptop.</a></li> <li><a href="https://tech.crescevo.com/microsoft-s-6502-basic-is-now-open-sourc/" rel="noopener noreferrer">Microsoft's 6502 BASIC is now Open Source (2025)</a></li> </ul> <p><em>Originally published on <a href="https://tech.crescevo.com/my-thoughts-on-the-bun-rust-rewrite/" rel="noopener noreferrer">Tech at Crescevo</a> — subscribe free for more.</em></p> programming devops webdev tech ironbyte-rgb Thu, 09 Jul 2026 19:00:15 +0000 https://dev.to/crescevo/new-dsls-need-great-docs-to-survive-llms-16jm https://dev.to/crescevo/new-dsls-need-great-docs-to-survive-llms-16jm <h2> TL;DR </h2> <ul> <li>A new DSL can survive in the era of LLMs with great documentation, marketing, and tooling.</li> <li>Advanced tooling around legacy languages has decreased hallucinations in LLMs.</li> <li>A robust language server and extensive documentation are necessary for a modern language.</li> <li>Creating an AGENTS.md file can help a new language play well with LLM agents.</li> </ul> <p>A new DSL can survive in the era of LLMs by focusing on great documentation, marketing, and tooling, as stated in the article "How a new DSL may survive in the era of LLMs" by William Cotton. The article highlights the importance of advanced tooling, such as type checkers and linters, in reducing hallucinations in LLMs. According to the article, a modern language needs a robust language server and extensive documentation to be viable.</p> <h2> What the data shows </h2> <p>The data shows that the volume of source code and advanced tooling around legacy languages have contributed to the decrease in hallucinations in LLMs. As stated in the article, "it's not just the mere volume of source code to train on. It's the advanced tooling around these languages." The article also mentions that "hallucinations are caught by the type checker before the code is even run." This suggests that the use of advanced tooling has improved the accuracy of LLMs.</p> <h2> What this means for ai readers </h2> <p>For AI readers, this means that a new DSL will need to have great documentation, marketing, and tooling to be viable. The article suggests that a new language will need a robust language server and extensive documentation with a good onboarding flow. Additionally, the article proposes creating an AGENTS.md file to help a new language play well with LLM agents. For example, the article mentions that the Web Pipe language uses a "webpipe init --codex" command to create an AGENTS.md file.</p> <h2> What to do right now </h2> <p>To increase the chances of a new DSL surviving in the era of LLMs, developers can start by creating great documentation and tooling for their language. This includes developing a robust language server and extensive documentation with a good onboarding flow. Additionally, developers can create an AGENTS.md file to help their language play well with LLM agents. The article also suggests that landing pages can be used to quickly get across the purpose and use case of a new language.</p> <h2> Bottom line </h2> <p>The bottom line is that a new DSL can survive in the era of LLMs with great documentation, marketing, and tooling. By focusing on these areas, developers can increase the chances of their language being viable in the era of LLMs. As stated in the article, a modern language needs a robust language server and extensive documentation to be viable.</p> <h2> Frequently asked questions </h2> <h3> Q: What is necessary for a new DSL to survive in the era of LLMs? </h3> <p>A new DSL needs great documentation, marketing, and tooling to survive in the era of LLMs.</p> <h3> Q: How can hallucinations in LLMs be reduced? </h3> <p>Advanced tooling, such as type checkers and linters, can reduce hallucinations in LLMs.</p> <h3> Q: What is the purpose of an AGENTS.md file? </h3> <p>An AGENTS.md file can help a new language play well with LLM agents.</p> <h3> Q: How can developers quickly get across the purpose and use case of a new language? </h3> <p>Developers can use landing pages to quickly get across the purpose and use case of a new language.</p> <h2> Sources </h2> <ul> <li><a href="https://www.williamcotton.com/articles/how-a-new-dsl-survives-in-the-era-of-llms?ref=ai.crescevo.com" rel="noopener noreferrer">https://www.williamcotton.com/articles/how-a-new-dsl-survives-in-the-era-of-llms</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://ai.crescevo.com/if-llms-have-human-like-attributes-then/" rel="noopener noreferrer">Age of Empires II matches LLMs in functional completeness</a></li> <li><a href="https://ai.crescevo.com/llms-will-replace-8-track-duplication-en/" rel="noopener noreferrer">LLMs replace humans in 8-track tape optimization</a></li> <li><a href="https://ai.crescevo.com/unified-controllable-and-faithful-text-t/" rel="noopener noreferrer">LLMs now control CAD generation with state-of-the-art faithfulness</a></li> </ul> <p><em>Originally published on <a href="https://ai.crescevo.com/how-a-new-dsl-may-survive-in-the-era-of/" rel="noopener noreferrer">AI at Crescevo</a> — subscribe free for more.</em></p> ai machinelearning llm programming ironbyte-rgb Wed, 08 Jul 2026 19:00:15 +0000 https://dev.to/crescevo/age-of-empires-ii-matches-llms-in-functional-completeness-57m https://dev.to/crescevo/age-of-empires-ii-matches-llms-in-functional-completeness-57m <h2> TL;DR </h2> <ul> <li>Adrian de Wynter's research paper "If LLMs Have Human-Like Attributes, Then So Does Age of Empires II" challenges the assumption of anthropomorphic attributes in large language models (LLMs).</li> <li>The paper proposes a 'null' assumption, where one assumes LLM non-uniqueness instead of assuming anthropomorphic attributes.</li> <li>The research shows that Age of Empires II is functionally- and Turing-complete, presenting similar attributes to LLMs.</li> <li>The paper was submitted on May 29, 2026, and last revised on June 11, 2026.</li> </ul> <p>Adrian de Wynter's paper presents a unique perspective on the attributes of LLMs, suggesting that they may not be as unique as previously thought. The research focuses on the idea that any entity in a sufficiently powerful substrate could present similar attributes. The paper proposes a new approach to understanding LLMs, one that assumes non-uniqueness instead of anthropomorphic attributes.</p> <h2> What the data shows </h2> <p>The data presented in the paper shows that Age of Empires II, a videogame, can be used to demonstrate the non-uniqueness of LLM attributes. By building and training a simple neural network on the game, the researchers were able to show that the game presents similar attributes to LLMs. The paper also discusses the idea that the interpretation of perceived behavior may change with the substrate, highlighting the importance of explicit measurement criteria.</p> <h2> What this means for ai readers </h2> <p>For AI readers, this research suggests that the attributes of LLMs may not be as unique as previously thought. The idea that any entity in a sufficiently powerful substrate could present similar attributes challenges the assumption of anthropomorphic attributes in LLMs. This could have significant implications for the development and understanding of AI systems, as it suggests that the focus should be on the substrate and measurement criteria rather than the attributes themselves.</p> <h2> What to do right now </h2> <p>Readers can start by exploring the paper and its findings, and considering the implications of the research for their own understanding of AI systems. The paper proposes a 'null' assumption, which could be a useful approach for those looking to develop and understand AI systems. Additionally, readers can consider the idea that the interpretation of perceived behavior may change with the substrate, and think about how this might impact their own work with AI systems.</p> <h2> Bottom line </h2> <p>The research presented in "If LLMs Have Human-Like Attributes, Then So Does Age of Empires II" challenges the assumption of anthropomorphic attributes in LLMs and presents a new approach to understanding AI systems. The paper shows that Age of Empires II is functionally- and Turing-complete, and that the attributes of LLMs may not be as unique as previously thought. By considering the implications of this research, readers can gain a deeper understanding of AI systems and the importance of explicit measurement criteria.</p> <h2> Frequently asked questions </h2> <h3> Q: What is the main argument of the paper? </h3> <p>The main argument of the paper is that the attributes of LLMs may not be as unique as previously thought, and that any entity in a sufficiently powerful substrate could present similar attributes.</p> <h3> Q: What is the 'null' assumption proposed in the paper? </h3> <p>The 'null' assumption proposed in the paper is the idea that one should assume LLM non-uniqueness instead of assuming anthropomorphic attributes when setting up an experiment.</p> <h3> Q: What is the significance of Age of Empires II in the paper? </h3> <p>Age of Empires II is used in the paper to demonstrate the non-uniqueness of LLM attributes, by building and training a simple neural network on the game and showing that it presents similar attributes to LLMs.</p> <h3> Q: What are the implications of the research for AI systems? </h3> <p>The research suggests that the focus should be on the substrate and measurement criteria rather than the attributes themselves, and that the interpretation of perceived behavior may change with the substrate.</p> <h2> Sources </h2> <ul> <li><a href="https://arxiv.org/abs/2605.31514?ref=ai.crescevo.com" rel="noopener noreferrer">https://arxiv.org/abs/2605.31514</a></li> </ul> <h2> Related reading </h2> <ul> <li><a href="https://ai.crescevo.com/llms-will-replace-8-track-duplication-en/" rel="noopener noreferrer">LLMs replace humans in 8-track tape optimization</a></li> <li><a href="https://ai.crescevo.com/unified-controllable-and-faithful-text-t/" rel="noopener noreferrer">LLMs now control CAD generation with state-of-the-art faithfulness</a></li> <li><a href="https://ai.crescevo.com/ponytail-make-your-ai-agent-think-like-t/" rel="noopener noreferrer">Ponytail – make your AI agent think like the laziest senior dev in the room</a></li> </ul> <p><em>Originally published on <a href="https://ai.crescevo.com/if-llms-have-human-like-attributes-then/" rel="noopener noreferrer">AI at Crescevo</a> — subscribe free for more.</em></p> ai machinelearning llm programming