| Project | What it is | Highlights |
|---|---|---|
field-agentFDE · RAG · evals |
Enterprise AI support-agent deployment kit — a RAG-grounded Claude agent with deterministic guardrails and an eval harness that scores every change in CI | 14/15 eval · CI-gated · $0 guardrail escalation |
torqueSDE · systems |
Distributed job queue for Node/TS — retries + backoff, dead-letter queue, visibility timeouts, crash recovery | ~1.2M jobs/s · in-mem + Redis · 90% cov |
hola-voicemailAI · voice |
Real-time AI voice assistant that answers, screens & transcribes calls in a natural voice | live MVP · GPT + ElevenLabs · WebSocket |
adaptive-book-learningAI · full-stack |
Turns any book into a personalized course — Claude builds a knowledge graph, then teaches & grades you | knowledge graph · spaced revision |
More → docuquery-rag (RAG) · typeahead-search (consistent hashing + WAL) · Drishti (accessibility AI, iQOO Hackathon) · lld-playground (system design)
I build and ship LLM-powered products end to end — voice agents, RAG systems, and agentic workflows (Claude Code, Codex) — plus the unglamorous parts that make them real: evals, latency budgets, and human-handoff design.
Currently Product & Tech Intern @ Krafton — helped build Swag.gg. Open to SDE · FDE · Product roles.


