The latest articles on DEV Community by Ahmed Khan (@ahmed_khan_6c5f55092f881b). https://dev.to/ahmed_khan_6c5f55092f881b 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%2F4061870%2Fa3b11883-6a31-494c-974a-65122647a5fa.jpg https://dev.to/ahmed_khan_6c5f55092f881b en Ahmed Khan Tue, 04 Aug 2026 06:58:08 +0000 https://dev.to/ahmed_khan_6c5f55092f881b/look-for-long-horizon-agents-for-frontier-labs-located-in-mountain-view-remote-3b5p https://dev.to/ahmed_khan_6c5f55092f881b/look-for-long-horizon-agents-for-frontier-labs-located-in-mountain-view-remote-3b5p <p>Bespoke Labs is looking for a researcher to help design and evaluate RL environments and benchmarks for long-horizon agentic tasks — the kind that take an agent hours, days, or weeks of coherent multi-step reasoning to complete, not single-turn prompts.</p> <p>What you'll do</p> <p>Design and build long-horizon RL environments and verifiers grounded in real-world tasks (code, tool-use, or enterprise workflows)<br> Develop evaluation benchmarks that measure agent coherence, planning, and reliability over extended trajectories<br> Analyze failure modes in long-horizon rollouts (drift, reward hacking, loss of task state) and propose fixes<br> Collaborate with the broader team on open datasets and reproducible eval recipes</p> <p>Must-have (hard requirement)</p> <p>Demonstrated long-horizon agent/RL experience — this is non-negotiable. You should be able to point to specific work involving multi-step, multi-day, or sequential-reasoning agent systems (e.g., contributions to environments like SWE-bench, Vending-Bench, FrontierSWE, DeepSWE, OpenReward, Gymnasium, or equivalent original research/production work). Applications without concrete long-horizon evidence will not be considered.</p> <p>Strong Python; comfort with RL training/eval frameworks (e.g., Verifiers, Gymnasium-style APIs, or custom environment tooling)</p> <p>Track record of publishing or shipping work others can verify (GitHub, papers, benchmarks, or production systems)</p> <p>Nice to have</p> <p>Experience with reward-hacking detection or "fuzzy" quality verifiers beyond pass/fail correctness<br> Background in multi-agent coordination or agent memory systems<br> Prior contributions to open-source RL environment or agent-eval projects</p> <p>Logistics</p> <p>Type: Contract, remote<br> Location: Remote (any timezone considered; some overlap with US/India hours preferred)<br> Compensation: Based on experience — happy to discuss</p> <p>How to apply</p> <p>Send a short note plus links to your relevant long-horizon work (GitHub, papers, benchmarks, or production systems you've shipped) to [<a href="https://experts.bespokelabs.ai/expert/apply/mts-long-horizon-coding-tasks-ER000018?src=JOsj6ol5" rel="noopener noreferrer">https://experts.bespokelabs.ai/expert/apply/mts-long-horizon-coding-tasks-ER000018?src=JOsj6ol5</a>]. No long-horizon evidence, no need to apply — we will reject on this filter first.</p> agents ai llm machinelearning