DEV Community: Anshuman Rai

HuskDesk: Stop Losing Your Flow to Tab Chaos | Week 3 Dev Challenge | Hacktoberfest 2026

Anshuman Rai — Thu, 22 Oct 2026 10:40:54 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady

HuskDesk — Close the Tabs. Keep the Context.

Your browser remembers the pages. But does it remember what you were doing?

We've all been there.

Twenty tabs open. Three unfinished projects. A documentation page you swear you'll read later. An important task buried somewhere between GitHub, Stack Overflow, and a dozen other tabs.

You close your browser to get a fresh start—and suddenly, getting back into your workflow feels like starting from scratch.

I wanted to solve that problem.

Meet HuskDesk: a local-first Chrome extension designed to help you organize your browser workspace, preserve your progress, and return to unfinished work without losing context.

The idea is simple: let your browser handle the clutter, so you can focus on the work that actually matters.

What I Built

HuskDesk is a productivity-focused Chrome extension built around one principle:

Your workflow shouldn't disappear just because you close your tabs.

Here's what it can do:

📂 Workspaces: Save collections of open tabs as named workspaces and restore them when you're ready to continue.
🌙 Tab Hibernation: Save and verify tab information before closing tabs to free up your workspace.
🧠 Resume Memory: Keep notes about what you were working on, what matters, and what you need to do next.
🗂️ Tab Organizer: Group tabs using domain-based organization, with experimental local AI-assisted grouping.
🔄 Pick Up Where You Left Off: Reopen saved workspaces and individual tabs without having to reconstruct your workflow manually.
🎨 A Focused Interface: Manage your browser work through a clean dashboard with light and dark themes.

HuskDesk isn't trying to make you work faster every second. It's trying to remove the little interruptions and repetitive tasks that make focused work harder.

Demo

GitHub Repository: github.com/hey-anshuman/HuskDesk

Huskdesk is currently distributed as an open-source project. You can explore the code, follow the setup instructions in the README, and build the extension locally.

Code

🔗 Explore HuskDesk on GitHub:

https://github.com/hey-anshuman/HuskDesk

The repository includes the extension source code and project documentation covering the architecture, design decisions, implementation status, security considerations, and test plan.

Built with a modern web stack, HuskDesk uses React, TypeScript, and Vite for its extension interface and Chrome's extension APIs for browser workspace management.

Contributions, feedback, and ideas for making everyday browser workflows less frustrating are welcome.

How I Built It

HuskDesk is built with React, TypeScript, Vite, and Chrome Extension Manifest V3.

The interesting part of this project is its approach to AI.

Instead of making a paid cloud AI API a requirement, I integrated Gemma 3 1B through Ollama, allowing the experimental AI grouping feature to run using a model on the user's own machine.

The architecture combines two approaches:

1. Reliable productivity features

Workspace capture and restoration, tab hibernation, and Resume Memory provide the core experience. These features don't require an AI account.

2. Optional local AI assistance

The experimental AI grouping workflow sends requests to a locally running Ollama service. The model can suggest how tabs should be grouped, while domain-based organization provides a fallback when AI grouping isn't available or produces unsuitable results.

This separation matters. The extension's fundamental purpose remains useful without requiring users to configure an AI model.

Local AI grouping is still experimental, and it requires users to install and run Ollama separately. It isn't a fully autonomous agent, and the quality of its suggestions can vary.

Why Does Open Innovation Matter?

For a productivity tool that works with someone's browser activity, the choice of AI architecture matters.

I wanted local AI to be an option, not another subscription.

Using an open-weight model with local inference offers several advantages:

More control: Users can run inference on their own machine instead of depending on a hosted AI API.
Privacy by design: The AI grouping request can stay on the local machine rather than sending tab information to a remote AI provider.
No mandatory paid AI API: Users don't need to purchase cloud model credits just to try the AI feature.
Room to experiment: Developers can inspect the integration, test different models, and improve the workflow without tying the project to one proprietary AI service.

Open innovation also makes the project easier for other developers to learn from, inspect, and extend.

Of course, local inference isn't automatically a guarantee that every part of an application is offline or that all stored data is encrypted. HuskDesk still needs to be transparent about its permissions, local storage, and optional AI setup.

For me, open-source software and open-weight AI aren't just technology choices. They're ways to give users more control over the tools they rely on every day.

The Lazy Payoff 🦥

The best productivity tool isn't necessarily the one that does the most things.

Sometimes, it's the one that saves you from doing the same small things over and over again.

With HuskDesk, the intended payoff is straightforward:

Spend less time reopening tabs and reconstructing a workspace.
Keep important context in Resume Memory instead of relying on memory alone.
Reduce tab clutter without losing the information needed to continue.
Experiment with local AI-assisted organization without making a paid AI service mandatory.

I haven't measured a reliable time-saving figure yet, so I won't claim that HuskDesk saves a specific number of minutes or hours. That's something I'd like to measure through real-world use and feedback.

The goal is to make returning to work feel less like rebuilding a desk after someone has cleared it—and more like sitting down exactly where you left off.

Close the tabs. Keep the context. Get back to what matters.

Prize Categories

Primary submission: Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady.

HuskDesk uses Gemma 3 1B through Ollama, so the Best Use of Gemma category may also be relevant, provided the implementation meets that category's official rules.

I'll only claim partner categories that HuskDesk genuinely qualifies for under the challenge guidelines.

Built with curiosity, local-first thinking, and a little less tab chaos.

🔗 Project: HuskDesk on GitHub

Feedback and contributions are welcome!

BazaarBrain - AI Powered Sales and commerce assistant | Week 2 Dev Challenge | Hacktoberfest 2026

Anshuman Rai — Sun, 18 Oct 2026 16:56:04 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

BazaarBrain is an AI-powered sales and commerce assistant designed to help local shop owners manage their business and create better marketing content — without needing any technical expertise.

Small retailers often manage products, inventory, customer information, and orders across notebooks, spreadsheets, and disconnected apps. Writing product descriptions, promotional messages, and maintaining a consistent brand voice takes extra time that shop owners would rather spend serving customers.

BazaarBrain brings these workflows together in one modern, commerce-focused dashboard.

The platform includes:

Dashboard & Analytics: An overview of business activity and key metrics.
Product Management: Add, edit, and delete products with form validation and SKU uniqueness checks.
Inventory Management: A dedicated interface for managing product stock levels.
Order Management: Tools for organizing and tracking customer orders.
Customer Management: A dedicated customer relationship interface.
AI Studio: Generate product descriptions, SEO content, WhatsApp promotions, Instagram captions, and FAQs — in English, Hindi (Devanagari), and Hinglish (Roman script).
Brand Memory: Maintain brand-related information to support consistent, on-brand marketing content.
Marketing & Automation: Interfaces for organizing marketing workflows and automation ideas.
Local-First AI: Built with a path toward running open-weight language models locally, reducing dependence on paid AI APIs.

BazaarBrain is designed for independent retailers, neighborhood shops, small business owners, and early-stage commerce businesses that need practical digital tools without the complexity of enterprise software.

Demo

🚀 Live Backend: bazaarbrain.onrender.com

The backend is live and deployed on Render. Frontend deployment and production integration are being finalized — a public demo link will be added once the core user flows are verified.

Code

hey-anshuman / BazaarBrain

BazaarBrain is built with a React + TypeScript frontend (Vite) and an Express + TypeScript backend, with PostgreSQL and Prisma as the data layer. The repository contains all application code, feature modules, backend services, and project documentation.

How I Built It

BazaarBrain is built around a modular full-stack architecture, with a focus on practical commerce workflows and open-weight AI.

Frontend

React + TypeScript
Vite
Modern, responsive SaaS-style interface

Backend

Node.js + Express 5 + TypeScript
Prisma ORM
PostgreSQL

Open-Weight AI (the heart of the project)

Ollama for local model inference
Qwen3 4B — primary content generation model
Llama 3.2 3B — lightweight fallback

The AI Studio supports several content-generation workflows: product descriptions, SEO metadata, WhatsApp marketing messages, Instagram captions, and FAQs. Language handling supports English, Hindi (Devanagari script), and Hinglish (Roman script) — making it genuinely useful for Indian local businesses.

A key design decision was to use local inference via open-weight models rather than making paid proprietary APIs a mandatory dependency. This keeps experimentation accessible and gives full control over how the AI component runs.

Why Does Open Innovation Matter?

Small businesses should not have to depend on expensive proprietary AI services to benefit from intelligent software.

Open-weight models and open-source tooling make it possible to:

Lower the barrier to experimentation — develop and test AI features without paying for every inference call.
Retain control — the model, prompts, and workflows can be adapted to different retailers' needs.
Grow freely — the modular architecture allows improving the AI component, adding new workflows, and exploring different deployment strategies as the project evolves.

For BazaarBrain specifically, local inference via Ollama means a shop owner (or a developer supporting them) can run the full AI Studio on commodity hardware. No API key, no monthly subscription, no vendor lock-in.

Open innovation isn't just about using an open model — it's about making the resulting technology more adaptable and giving developers the freedom to build genuinely useful products around it.

My Agent Session

Agent-assisted development was a core part of how BazaarBrain was built — covering frontend architecture, backend integration, debugging, and deployment configuration.

A DevRelay agent session link will be embedded here once a shareable session is available.

Prize Categories

🤖 Open-Source AI: BazaarBrain uses open-weight models (Qwen3 4B and Llama 3.2 3B) via Ollama for local inference — no proprietary AI API required.
💼 Side Hustle: BazaarBrain is a practical AI sales assistant for local shops, with a clear path toward a useful, monetizable product for small businesses.

DEVFEST 2024

Anshuman Rai — Sun, 09 Feb 2025 20:46:00 +0000

Venue : UIT Auditorium
Date : 26 November 2024
Organiser : GDG Prayagraj

Devfest...!!
What is devfest? What is the first thing that clicks in your mind when you hear this? Well when I was unknown about it I thought it's any cultural fest that is organised by GDG but when I entered into the world of tech and developer I realised devfest is not just only about bringing culture but it is also about bringing tech community and developer together to celebrate and learn more about the recent trends in technology and to discuss different aspects of tech in various fields.

Throwback to the day when I Just got back from DevFest 2024 by GDG Prayagraj — and wow, what an experience!

The fest started around 10am. We got our ID cards and GDG wrist sticker band. This was a feeling of getting into a concert which more hyped the excitement me and my friend as an attendees. We entered the Auditorium after getting official passes from counter where the gdg core team perfectly managed to response each registration. The session started with keynote of GDG organiser Ankit Kumar Verma Sir. And went on with different speakers in row.

From learning about the future of Generative AI to seeing AI-powered mobile apps built in Flutter - Every session had something fresh to offer. The speakers? Phenomenal. People like Nishant Welpulwar, Dhrumil Shah, and Praveen Das sparked endless ideas and inspired new ways to look at technology.

Leave it to DevFest to turn ‘Ctrl + Alt + Del’ days into ‘Ctrl + Fun + Code’ moments!

A few highlights:

Trust and Safety: Insights from Google’s Vivek Kumar on building secure, inclusive spaces.

Serverless AI: Anubhav Singh’s deep dive into Cloud Run and GPU was next-level.

Open Source Innovations: How Praveen Das uses tech at Wikimedia to drive collaboration.

IDX and Community: Ankit Kumar Verma 's keynote on responsible AI and community-building.

This was my very first time when I was attending Devfest. I was very much excited to meet industry experts and to know their valuable insights. For me Devfest is much more than opportunities because it is the complete blend of learning, innovation, collaboration, network and technology. It was privilege to be there at Devfest 2024.

If all conferences were like DevFest, I’d be attending more ‘tech talks’ than family dinners! Big thanks for making code cool !
This the the whole expression and experience that I can quote through my this statement.

As the end approached,the atmosphere grew even more intense and exiting. Attendees had lot of fun and food . And I must mention this GDG's hospitality was outstanding and foods were really delicious. They took care of our needs like food and hygine. The Devfest speakers were really fabulous and best in their respective fields but I personally enjoyed the session of Dhrumil Shah , the way he was interacting with audience was really great and fantastic.

After all the sessions, the moment we were all eagerly waiting for finally arrived—GDG Swags! Our goodies included a diary, eco-friendly pen, GDG & TFUG stickers, and a carry bag. We clicked countless photographs and created unforgettable memories filled with fun and learning.

Grateful to the GDG Prayagraj and team for making it all happen! Connecting with fellow tech enthusiasts and creators was priceless. Already looking forward to putting these insights to use—and staying engaged with this amazing community.