leveraging multi-agent reinforcement learning (marl) to enhance e-amod fleet control
we built on an exisiting tri-level framework that utilizes graph-rl to solve the e-amod fleet control problem. we test how well a marl approach performs given a situation where vehicles in an e-amod fleet need to be continuously recharged and spatially rebalanced.
tuning video segmentation for creating masks of pedestrians
we used fair’s recent segment anything model (sam) to create masks of pedestrians from driving footage. fine-tuning sam for pedestrians can lead to easier applications of machine learning models for pedestrian agent prediction that would otherwise be too computationally expensive to train on video data
stanford treehacks 2023
project: condensis
my team and i built condensis in 48 hours - our web-app utilizes openAI's api to generate condensed lecture notes from videos of college lectures - the multimodal language model powering it can synthesize high-quality, easy-to-understand notes from any lecture video