ML Engineer · San Jose, CA
I build the stuff between the paper and the product.
On-device models, agent systems, things people actually use.
* I'm Shyam. I train models and then make them do something useful. Most of my time goes into closing the gap between research and real users. I own the whole pipeline. .deploy() from training runs to prod, and I don't consider it done until someone's using it.
Selected projects
Ridesharing for SJSU students. On-device ML: A* pathfinding, Vision OCR for student IDs. No images ever hit a server.
Point it at a repo, it reads everything, writes the docs, and opens a PR. Four agents working in sequence.
Ask a biomedical question, it pulls live PubMed papers and generates a grounded answer. BioGPT + RAG.
*I like the messy middle, where the model works in a notebook but not in prod yet. That's where I live.
The stuff that runs when no one's watching. Training loops, quantized models on phones, inference that doesn't crash at 2am.
Wiring up agents that actually do things: read repos, pull papers, open PRs. Not chatbots. Systems.
If the product needs an iOS app I'll write Swift. If it needs infra I'll set up the cluster. I go where the bottleneck is.
M.S. Artificial Intelligence
San Jose State University
B.S. Computer Science
Arizona State University · 2023
I got into ML because I wanted to build things that felt like magic. Stayed because the hard part isn't the model, it's getting it to work for someone who doesn't care how it works.
San Jose, CA
M.S. AI @ San Jose State
Lifting · Photography · Music
Got a hard problem or an ambitious idea? I'm in.
