At Walmart, I build backend services for recommendation serving and personalized discovery. I built a real-time retrieval path using EmbeddingGemma embeddings and filtered ANN search, and resolved Milvus client timeouts in a way that increased complementary-item recommendation coverage by 30%.
I developed Local Marketplace personalization services and seller-store recommendation pipelines, contributing to $5M GMV. I also launched local-seller recommendations on Post Add-to-Cart and developed its Fashion journey, which delivered 43.1% GMV growth and 27.2% year-over-year page-visit growth for that journey.
At UCF Smart & Safe Transportation Lab, I designed a fisheye-camera surveillance system for vehicle counting and path-data collection, and integrated YOLOv4 and Faster R-CNN detectors with CSRT and KCF trackers. That research contributed to a publication on detecting and tracking vehicles in fisheye traffic videos.

