Work Experience

• Reduced search latency from 2s to under 500ms across semantic + fuzzy queries, by redesigning PostgreSQL indexing strategy and building a custom search engine around it.

• Cut LLM API costs and improve extraction accuracy, by replacing Gemini-based OCR parsing with a custom regex parser — removing the dependency entirely.

• Improved data pipeline reliability to ingest 25,000 tender records/day into production PostgreSQL, by building and operating fault-tolerant Python scrapers across multiple portals.

• Enabled targeted outreach at scale by designing an automated WhatsApp delivery system using queues and background workers with dynamic business rules.

• Built and optimized AI image generation pipelines using Next.js frontend and Node.js/Express backend, improving latency by 20%.

• Shipped full-stack features across React.js frontend and Express/PostgreSQL backend, owning modules from implementation through release.

• Built and maintained Node.js & Express backend services and PostgreSQL databases.

• Delivered new features and resolved critical bugs to improve product stability.