ML systems engineer. Google DeepMind GSoC 2025 alumnus.
I build infrastructure that makes AI practical at scale. My GSoC project—the Gemini Batch Prediction Framework—reduced API costs 75% through async batching and context caching.
Currently pursuing post-bacc CS/Math, preparing for doctoral research in reinforcement learning and efficient ML systems.
Gemini Batch Prediction Framework — Production async pipeline for Gemini API. Intelligent batching, context caching, 95%+ test coverage. GSoC 2025 with Google DeepMind.
ContextRAG — Adaptive retrieval processing for LLMs. 30% token reduction via document-length-aware chunking.
paperweight — Academic paper filtering. 300+ papers/hour, 85% relevance precision.