Leading PyTorch and Triton GPU performance, optimizing e2e model training and inference.

Previously worked on Meta’s deep learning recommendation models, Triton, and PyTorch at Meta SuperIntelligence Labs.

Former maintainer of the TorchRec library: https://github.com/pytorch/torchrec

Assisted undergraduate and graduate students in Java and Android development. Worked on a Google Scholar Analytics Service

Designed and developed REST APIs in Java for programmatic access to data on an ETL tool platform, reaching hundreds of customers and resulting in a 25% decrease in latency for average customer use. Went above and beyond by implementing a complex log filtering system using recursive descent query parsing.

Designed and engineered a full stack project on a tier-1 service for automating alerts and alarms to 1000+ AWS customers, resulting in ~30% faster response times from support engineers.