Deepak Kumar · Applied AI · LLM & agent engineer

I build LLM agents that do real work, and write about how they actually ship.

I design and ship agentic AI systems end to end, from the prompt and eval loop to the infrastructure around it. I run ToolJunction, an AI discovery platform, and build the autonomous systems that run the business behind it.

I also write the honest version of how these systems get built: deep technical build logs that engineers actually read. Open to applied-AI / LLM / agent-engineering roles and select technical-writing engagements.

22K+ GitHub stars
1,100+Followers
90+Open-source repos
2016Building since
Side projects

The things I build outside my day job, taken further than a weekend hack.

Open source
Writing

I write deep, honest engineering build logs about shipping AI systems, the architecture, the trade-offs, and the bugs nobody blogs about. Available for sponsored technical deep-dives and developer education.

Jul 2026 · Build log · 14 min
I built an AI agent that handles 1000+ sales inquiries a month
The architecture, the six bugs that mattered, and the money an AI audit found in threads everyone had given up on. A full walk through of designing agentic systems with real guardrails and human-in-the-loop review.
What I work with
AI
LLM agents & tool use, prompt & eval design, RAG, structured decisioning, human-in-the-loop systems
Backend
Python, Node / Express, SQLite & MongoDB, REST APIs, email & sync infrastructure (IMAP/SMTP)
Frontend
Next.js, React, TypeScript, vanilla JS when it is the right tool
Ship
Cloudflare, Vercel, launchd/cron, prototype to production, solo
Let's work together

Hiring for applied AI, LLM, or agent engineering? I ship agentic systems end to end and would love to talk.
Need a technical deep-dive written? I take on a few sponsored build logs and developer-education pieces each month.

[email protected]