I build AI agents and publish the code.

Most projects start the same way: I run into an interesting problem, build the smallest useful version, test it end to end, and add it to Awesome LLM Apps.

The repo now has more than 100 open-source projects and 136K+ stars. Some are small enough to understand in one file. Others are full agent teams with tools, memory, voice, browsers, RAG, and interactive UIs.

Lately, I've been working on

• Agent Skills for Claude Code, Codex, Cursor, and other coding agents
• Always-on agents that monitor things and do useful work in the background
• Multi-agent teams for research, finance, coding, design, and other real workflows
• Voice agents, browser agents, MCP apps, and generative UI
• RAG, memory, evals, tool calling, and the less exciting parts that make agents reliable
• Full code-first crash course on Agent Development Kit and OpenAI SDK

Everything is built to be cloned, run, pulled apart, and adapted. The projects work across Gemini, Claude, GPT, DeepSeek, Llama, Qwen, and other open-source models.

• Senior AI Product Manager at Google Cloud helping developers BUILD, SCALE & GOVERN AI Agents with ADK, Agent Builder, Agent Engine and Vertex AI platform.

• Co-author of "GPT-3: The Ultimate Guide To Building NLP Products With OpenAI API" and "Neural Search - From Prototype to Production with Jina"

My wife Gargi and I run Unwind AI, where we share the AI tools, tutorials, and news we find genuinely useful.

I also invest in early-stage AI agent startups, usually with checks between $25K and $50K.

If you're building something interesting with agents, feel free to reach out.

Or just pick a project from the repo and start breaking it. That's usually more fun.