Namra Patel

Hello! I'm a Master's student at Georgia Tech advised by Animesh Garg. I'm interested in robot learning from human video. More specifically, I want to understand what makes human video transfer to robot action, and how to encourage it. To this end, I work both on developing methods for learning from human video and on studying the transfer itself. Previously I worked on distributed systems such as Ethereum's ERC-4337, and on high-performance game servers at Argus.

Email / X / Github

Research

FloMo: 3D Scene Flow as a World-Action Model Intermediate for Learning from Human Video Jeremy A. Collins*, Namra Patel*, Ayush Agarwal*, Shitij Govil*, Animesh Garg Under review

A 5B video-generation model trained to predict 3D scene flow and robot actions. Rendering scene flow as RGB lets the pretrained video VAE encode it directly, and predicting this motion rather than future video is far more sample-efficient for learning actions.

Quantifying the Generalization Value of Pretraining Tasks in Robotic Manipulation Undergraduate Thesis, Western University, 2025

A framework for quantifying the transfer value of a robotic manipulation task to other tasks in a taxonomy. Shows that both foundational tasks and complex tasks accelerate learning on others, but only tasks higher in the hierarchy can be used to improve performance on simpler tasks.

Writing

July 14th, 2025"Convolutional" layer backpropagation, done right

June 24th, 2025An Analytical Approach to Batch Normalization Gradients

June 17th, 2025Policy Gradient Methods

June 10th, 2025Value Function Approximation

May 27th, 2025Model-free Control

May 22nd, 2025Model-free Prediction

May 18th, 2025Policy and Value Iteration

Template