Parag Mandal
Staff Engineer
I build systems that make data less chaotic and AI slightly less confused. My work lives at the intersection of RAG, NLP, and distributed pipelines—where things either scale beautifully or fail spectacularly. Currently leveling up in ML at Georgia Tech while trying not to overfit my own expectations. I care about performance, clean abstractions, and solving problems that actually matter. Bonus points if it runs fast and breaks rarely.
EXPERIENCE
- ›Intelligence platform for market and customer data with Row-Level & Attribute-Level Security via React, FastAPI, and LangChain RAG
- ›Multi-agent AI deployments using Model Context Protocol (MCP) servers integrated with Databricks Genie
- ›Enterprise Data Quality framework across Bronze / Silver / Gold medallion architecture
- ›Enterprise-wide metadata cataloging system spanning business applications and datasets using custom ingestion pipelines
- ›Internal Data Catalog platform on Databricks with automated schema discovery and lineage tracking
- ›Mapped enterprise ingestion pipelines from financial source systems with API catalog for data product discovery
SKILLS
PROJECTS
Computer vision pipeline with ResNet-50 backbone for human detection, tracking, and re-identification under occlusion.
Real-time wildfire detection with YOLOv8 + SAHI for improved small-object accuracy in complex terrain.
Health assistant aggregating trusted medical sources and answering queries via GPT-3.5 Turbo.
Blind-spot detection prototype using custom-trained YOLOv8 with real-time alerting for driver assistance.