Note: I’ve lost access to my original GitHub account (aayush1205), so I’ve forked my older repositories here for continuity.
I'm a second-year Computer Science MS CS student at Texas A&M, currently a Research Assistant in the Department of Oceanography working on ML for physical sciences — using spatiotemporal Vision Transformers to predict heat transport across the Atlantic. It’s compute-heavy, high-dimensional, and way too fun.
Before grad school, I engineered distributed infra at LinkedIn, built services that touched 100K+ QPS, and designed fault-tolerant systems for ML production. I’ve also dabbled in GNNs, launched a D2C brand, and built a CLI music player because browsers were too much effort.
- ML Infrastructure & Distributed Training
- Ocean Modeling and Spatiotemporal Transformers
- RL for LLM Context Management
- 3D Computer Vision (Gaussian Splatting, NeRF, vSLAM)
- Representation Learning & Model Compression
- Reinforcement Learning for LLM Memory Efficiency
- Optimization
- Ocean Circulation Patterns via ML
