Kaustubh Ponkshe

I am a PhD student in Computer Science (EDIC) at EPFL, Switzerland, where I am part of the Machine Learning and Optimization Lab (MLO) led by Prof. Martin Jaggi.
My research explores the foundations and frontier challenges of modern machine learning—spanning large-scale optimization, foundation-model adaptation, and the reliability and safety of intelligent systems.

Earlier, I worked at the CERT-Lab at MIT and the MBZUAI with Prof. Praneeth Vepakomma, where I developed methods for efficient and federated fine-tuning of large models, studied alignment robustness and memorization in LLMs**, and contributed to research on privacy-preserving and distributed learning.

I have also been involved in the Apertus LLM initiative under the Swiss AI Initiative, contributing to its post-training and safety alignment efforts.

Before that, I completed my B.Tech in Electrical Engineering and M.Tech in Artificial Intelligence at IIT Bombay, where I led the Path Planning and Controls subdivisions of IIT Bombay Racing, building India’s first driverless racecar.

Outside research, I enjoy traveling (24 countries so far 🌍) and playing tennis 🎾.

Feel free to explore my publications or view the full list on my Google Scholar.