Advaith Ravishankar

Advaith Ravishankar is a Multimodal AI researcher at Harvard University, focusing on Multimodal Foundation models, World Models, Robotics, and Digital Twins. He is an Incoming Research Scientist at Applovin, focusing on Multimodal Content Understanding with Foundation Models and World Models.

Previously, he conducted research at the Salk Institute for Biological Studies at the Talmo Lab with mentorship from Dr. Talmo Pereira, developing sample-efficient and low-latency computer vision models for 2D/3D markerless tracking. He has also been mentored by Dr. Rajeev Jain at Qualcomm, working on generative AI methods for VLSI. He is currently a research assistant at the Harvard AI and Robotics Lab, working with Dr. Mengyu Wang.

He obtained his Master's Degree in Computational Science and Engineering at Harvard University, and he completed his undergraduate degree at UC San Diego, double majoring in Mathematics-Computer Science and Cognitive Science Specializing in Machine Learning and Neural Computation. For teaching, he was a Teaching Fellow at Harvard for Modern AI Systems: Agents and LLM System Optimizations (CS2860) and AI in Medicine (SHBT 261), teaching how to design multi-agent systems, deploy LLMs for Inference, finetuning LLMs with RLHF and buidling VLMs for Medicine.

Beyond research, he served as President of the Data Science Student Society at UC San Diego, one of the largest Data Science student organizations in the world, for the 2024-2025 term. In 2024, he served as a hackathon judge for DataHacks and was a public speaker at UC San Diego, delivering talks on Generative AI and providing research mentorship to student teams. At Harvard, he is a Kempner Research Mentor for the 2026 class under Multimodal and Embodied AI.