At SohamGlobal Group, I was the sole developer on an Intelligent Visitor Tracking System, shipping it end-to-end to address manual gate-entry logging. I engineered an OCR and fuzzy-matching pipeline that reached 95–96% end-to-end accuracy while keeping latency at 1–2 seconds per vehicle.
I designed normalized MySQL schemas and shipped an admin dashboard for analytics, access control, and suspicious-activity alerts. I also added a Linear Regression model to forecast next-day traffic.
At Altair, I cleaned and preprocessed multi-source datasets for an analytics pipeline. I also ran exploratory data analysis and built visualizations to present insights to leadership stakeholders.
In my projects, I built QueryMind, a self-correcting natural-language-to-SQL agent that reached a 92% query success rate without human intervention. I'm also developing TaskForge, a PostgreSQL-backed distributed job scheduler with multi-worker coordination and recovery for crashed workers.
