At Oracle, I develop Java microservices and PL/SQL for high-availability enterprise payment systems. I also manage database changes with Liquibase and support production services.
As a Research Assistant on Oracle’s PGX Team, I implemented GPU-based HNSW graph construction in C and C++ with NVIDIA CAGRA. It achieved roughly 8–9× faster index construction than the 8-thread parallel CPU baseline, with equivalent search time and recall.
I built reproducible Python benchmarking tools to measure latency, recall, and scalability, and presented results that informed the team’s indexing roadmap. I also designed a hybrid approach with GPU graph construction and CPU approximate nearest-neighbour search.
My personal projects include an HTTP/1.1 server, a Unix shell, a virtual machine, and a ray tracer, built from scratch in C and C++. My background in mechatronics and industrial automation also includes PLCs, microcontrollers, and electronics.

