raghav punnam
robotics @ harvard, berkeley eecs
about:
I'm a gradudate student at Harvard working on robotic foundation models for humanoid loco-manipulation. I'm interested in how human intuition and movement can inform robotic reasoning and generalize across different embodiments. Outside of research, I enjoy playing basketball and watching movies.
currently:
- - ms Applied Math & CS @ harvard
- - robot learning @ wang lab (prof. mengyu wang)
previously:
- - bs EECS @ UC berkeley (4.00 GPA)
- - multi-agent RL @ BLISS lab (prof. kannan ramchandran)
- - intern @ meta, state farm
- - teaching assistant @ berkeley EECS (CS170, CS161)
research:
- Multi-Agent RL for Vendor Pricing in Competitive Marketplaces — BLISS lab, MARL
- Adaptive Optical Compression for DeepSeek-OCR — CNNs, computer vision
projects:
- bubll →
food recommendation iOS app using NLP and group rec systems
- figpaper →
cursor-like web editor for LaTeX with AI code generation
- NYC citibike prediction model →
markov process + flow models over 15M+ trips and 2,000+ stations
- cross-domain recommendations w/ wasserstein alignment →
optimal transport, rec systems
- parameter criticality in biological vs. artificial neural networks →
neurobiophysics, RL