AI Engineer building agentic developer tools.
Full-stack engineer @ Shipsy · Gurgaon, India
I ship production systems for 100+ enterprise clients at Shipsy, where I work across the stack on TypeScript, NestJS, React, and MongoDB. Lately I've been going deep on LLM tooling and coding agents — the kind of infrastructure that makes other engineers faster.
I'm most at home where backend reliability meets AI ergonomics: webhook architectures, MCP servers, retrieval over codebases, and agent loops that actually hold up under real workloads.
Currently: open to AI Engineer roles at startups working on developer tools and coding agents.
| Project | What it does | Stack |
|---|---|---|
| archaeologist-mcp | Time-aware code context (git blame + history) for AI coding agents — surfaces why code looks the way it does, not just what's there. | Python, MCP |
| career-ops | AI-powered job-search system built on Claude Code with skill modes and batch processing. | JavaScript, Claude API |
| bolna-recruit-screen | Voice-agent recruitment screening with a realtime dashboard. | Next.js 15, Supabase |
| CardWise | Recommendation engine scoring spending profiles against 100+ credit cards — top-5 results in <2s, Lighthouse 95+. | React, TS, Node |
| Flee Outpass Portal | Biometric campus-access system live at IIIT Kottayam, serving 2,000+ students. | TS, React, Node, MongoDB |
- 🤖 Shipped an LLM-driven auto-config bot that compressed release cycles from weeks → days
- ⚡ Built automation reducing manual ops by 90% across 100+ enterprise accounts
- 🔁 Re-architected webhook infrastructure → 60% fewer integration failures, 99.9% sync reliability
- 🔐 Delivered enterprise security features (RBAC / SSO / 2FA) with zero post-launch incidents
- Daily — TypeScript · NestJS · React · MongoDB · Node.js · Redis · Docker
- LLM / AI — Claude API · MCP · Llama 2 (prod) · prompt eval pipelines · agent loops
- Comfortable in — Next.js · PostgreSQL · AWS · GCP · Python · Jest / Cypress
If you're building developer tools, coding agents, or LLM infrastructure — I'd love to chat.