Full-stack engineer who loves building things and learning something new along the way.
Distributed video processing
Video-processing workers coordinated through Raft consensus, with jobs queued on RabbitMQ and output kept in MinIO object storage. Most of the work went into the unhappy path: leader election, recovery, and keeping processing observable while a node is down.
Coding judge with model feedback
Runs submitted code in isolated sandboxes, then goes past pass/fail with complexity analysis and optimization feedback from a language model.
ML benchmarks in the browser
Hand-written WGSL kernels for machine-learning operations such as matrix multiplication and softmax, benchmarked on whatever GPU the browser has access to.
Built reservation and marketplace systems used across 10+ brands, with high-concurrency booking flows in Next.js, NestJS, and MongoDB.
Reworked regex-heavy network-log parsing as an LLM chunking prototype, then validated accuracy and latency against analyst workflows.
Built a Wasserstein GAN that synthesized 12-lead ECG signals from 2-lead KardiaMobile data with 95% similarity. Advised by Dr. Naveena Yanamala and Dr. Raj Reddy.
Improved cross-domain few-shot learning accuracy by 2% on fine-grained datasets using prompt-based contextual learning with CLIP. Advised by Dr. Soma Biswas.
M.S. Computer Science, University of Massachusetts Amherst
B.E. Computer Science, SSN College of Engineering · Anna University
If you’re working on distributed systems, applied ML, or developer infrastructure, I’d like to hear about it. Email is the best way to reach me.
