Jul 2024 – Present · 1 yr 8 mos · Bengaluru, Karnataka, India

• As a Software Development Engineer on the FCC Pricing team, I engineered and scaled critical flows using Python. My work focused on integrating the Milvus vector database to modernize legacy systems, leading to significant improvements in performance, cost savings, and operational efficiency for FCC-Pricing product matching.
• Accelerated a core product matching pipeline by over 12x, slashing processing time from over an hour to under 5 minutes. This was achieved by re-architecting the legacy system to integrate Milvus for high-speed vector similarity search, which also boosted recommendation upload throughput by nearly 10x.
• Engineered and automated a scalable Bestseller matching system to compare 20 million products, directly increasing revenue and saving a lot of infrastructure costs by leveraging efficient vector search with Milvus and DiskANN.
• Led the end-to-end production hardening of the Milvus vector database infrastructure, establishing it as a core, reliable service by implementing automated backup/recovery strategies, custom monitoring scripts, and robust log management.
• Developed a fully automated, DAG-based workflow and for the Bestseller matching process. This empowered operations teams to run jobs independently, shifting the process from a manual effort to a highly efficient, automated system.
• Ensured high system reliability by resolving critical production bugs during a high-traffic period maintaining core functionality under pressure

Jan 2024 – Jun 2024 · 5 mos · Bengaluru, Karnataka, India