I build AI agents. By day: production agents that grade, moderate, and analyse at scale, plus the GPU infrastructure that serves them. On the side: research on getting small models to act (SFT and GRPO in a live browser) and on coding agents that rebuild programs from stripped binaries.
Plain-text summary of this profile, regenerated daily
- Who: Naman Chawla, AI/ML engineer in New Delhi, India
- Focus: building and training AI agents; LLM post-training (SFT, GRPO), quantization, distillation; agent-assisted reverse engineering
- Production work: LLM agents and APIs for grading, content moderation, analytics, and proctoring (private repositories)
- Public research code: openEnvRL, capl, quake2-llm-decompile
- Full machine-readable index: llms.txt
- Updated: 2026-11-03, covering the last 12 months, public and private work
- Contributions: 1083; commits: 920; pull requests: 127; repositories owned: 53
- Streak: current 0 days, longest 7 days
- Busiest weekday: Thursday; peak commit hour: 00:00 IST
- Languages by code size: Python 77%, JavaScript 13%, TeX 3%, Shell 2%
Charts above the snake are regenerated daily by gen.py and include private work as counts only.