At Deloitte, I contributed to developing and supporting an enterprise LLM platform, including its RAG functionality, testing, and debugging. I implemented persistent chat history backed by SQL and usage guardrails for controlled use.
For Disney, I profiled and optimized a Python video-similarity pipeline, contributing to GPU acceleration with CUDA, CV-CUDA, and FFmpeg. The work reduced runtime by approximately 78%, with consistency validated between CPU and GPU results.
At OpTeamizer, acquired by Deloitte, I developed and supported Embedded Linux software for medical-device and hardware-integrated systems. I also built a local LLM/RAG server with FastAPI, Docker, and Milvus, including a LangGraph document-comparison agent with source-grounded results.

