Madhav Kanda

I am a Senior Associate, Applied Research at Capital One in San Jose. I work on inference optimization and efficient training.

Previously, I completed research internships at Microsoft Research, Aalto University, and the Indian Institute of Science (IISc), Bengaluru.

I completed a fully funded M.S. in Computer Science at the University of Illinois Urbana-Champaign. I earned my B.Tech. with Honours in Computer Science from IIT Gandhinagar. At IIT Gandhinagar, I received the Gold Medal for Outstanding Performance and was named to the Dean’s List every semester.

My research interests lie in efficient and reliable generative AI. I currently explore diffusion models, distillation, and speculative decoding. My goal is to make language models more practical for agentic workflows and synthetic data generation.

My earlier research spans multi-agent workflow evaluation, probabilistic program synthesis, Bayesian inference, and multimodal reasoning. Across these areas, I aim to build AI systems that reason reliably and use computational resources efficiently.