Working on scalable distributed training of large language models using SageMaker’s model parallelism and communication libraries. Focused on optimizing training across multi-GPU and multi-node AWS infrastructure.

Awarded the Undergraduate Teaching & Research Award (UTRA) to collaborate on advancing 3D computer vision research under Professor Srinath Sridhar. Currently trying to make 4D Gaussian Splatting faster by leveraging VLMs.

Served as a teaching assistant for CSCI 1230: Computer Graphics in the Fall 2024 semester, the longest running graphics course in the known universe.

Contributed to the development of an adaptive augmented reality labeling tool, enhancing skills in AR development and image analysis. Developed and deployed a Unity-based AR software suite for navigating over 20 realistic testing environments. Conducted a user study with 18 participants to enable real-time interaction and evaluation of AR labeling models. See paper here.

Served as a teaching assistant for CSCI 1430: Computer Vision in the Spring 2024 semester. Host weekly office hours to support students with coursework, clarifying complex concepts and methods in visual computing. Collaborate with course staff to develop and update course materials, including lectures, project guidelines, and grading tools.

WorldMAR is a fast, action-conditioned next-frame generator for Minecraft, inspired by Oasis and the MAR framework for image generation. Designed for model-based control, it enables ~4× faster sampling than full-frame diffusion methods.

LeCoin is an educational cryptocurrency inspired by Bitcoin, featuring proof-of-work mining, transaction verification, and fork resolution. It simplifies blockchain mechanics using global balance tracking instead of UTXOs, and runs over a custom virtual OS and peer-to-peer network for decentralized coordination.

A from-scratch stable diffusion model for conditionally generating street view images from a corresponding satellite image. Given a satellite image, the model is able to generate a plausible looking view from the ground.

A simulation modeling the transport networks generated by Physarum (a slime mold). Implemented with Unity and powered by a compute shader, capable of individually updating the dynamics of 1 million+ interacting slime agents in real time.

A better Word Hunt! Developed with React and playable here! Click tiles to select them and type to place a letter. After creating your desired shape, press randomize to switch things up and drag together adjacent letters to form words. Longer words = more points!

A set of models for predicting country and coordinates from a given street view image. Scraped 50,000+ images from Europe using Selenium + Google StreetView API, and separately fine-tuned ConvNeXt and BiT for country and real-valued coordinate predicitions. Achieved 61% country classification accuracy (seems low, but you try it with a single street view image!) and 60% accuracy on coordinate prediction within 500km.