I like to talk about things that are interesting or important to me. Most of them are related to AI Systems' safety and performance.
I currently work at Secure AI Futures Lab on technical AI governance.
I have previously had the pleasure of working at SteadRise helping with all things technical in building capacity for AI Safety and Security. In May 2025, I completed my masters program in Computer Science at the Chennai Mathematical Institute. I previously worked part-time helping the SteadRise team design and manage technical evaluations and training to run the Global AI Safety Fellowship. Before that, I built dangerous capability evaluations of LLM agents for the AI Security Institute, UK as an external collaborator. Prior to this, I was a research fellow in SteadRise's Summer Research Fellowship.
In my past life, I was a Software Engineer at Microsoft.
Links
- GitHub
- LeetGPU
- AutoEnhance Announcement
- BlueDot's AI Safety Fundamentals - Notes
- Lambda Calculus - Notes
Open Source Contributions
- Inspect Evals — Ported the implementation of Humanity's Last Exam to Inspect AI Evals repository.
- NEKO — Enabled eval.py to work on text datasets.
- Modular / Mojo — Added simple but essential Leaky ReLU Activation kernel.
- Artificial Intelligence - A Modern Approach — Fixed EightPuzzle class implementation in search.py.