Syed Muhammad Hanzala

At Özyeğin University, I led the annual-filings RAG workstream for a retail research chatbot covering 190 companies. I built a Python pipeline over 158,570 SEC 10-K chunks, using PostgreSQL and pgvector retrieval to support citation-grounded answers.

I investigated skipped PDF content and recurring parsing failures by combining parsers with OCR, comparing recovered text against source documents, and rerunning validation checks. I also resolved out-of-memory failures on Ubuntu EC2 by processing pages in batches and clearing stale caches.

At BookJane, I built and deployed an internal leave-management system with Next.js, TypeScript, PostgreSQL, and Prisma. It included Google sign-in, role-based permissions, approval workflows, and email notifications.

On Echo, an encrypted LAN chat project, I built a real-time messenger with Spring Boot WebSockets and React, integrating five Python/FastAPI services for language and message features. I containerized the application and services with Docker Compose and automated startup and connections with Bash scripts.