Adeel Ahmed Qureshi

Answers carrier and state eligibility questions from internal documents, then acts on them — issuing policies, endorsements and cancellations inside the EPIC CRM, and generating Certificates of Insurance and ID cards through 80+ SOAP tool calls.

Writes aren’t blind. Each tool enforces its own constraints, and a separate validation agent checks every proposed action before it commits — so a wrong retrieval doesn’t become an issued document. A data team spot-checks output in production on top of that.

Each service rep went from managing two producers to ten, absorbing the team’s growth without an additional service hire — roughly $5,000 a month in avoided headcount.

Backend services for the insurance routing module embedded directly in the Trucker Path and Truck Loads apps, moving insurance leads out of the consumer products and into the sales pipeline.

Runs in production behind a product surface with more than a million users.

Led the move of a core microservice off AWS SAM onto a Kubernetes architecture on EKS, removing cold starts, wrapper overhead and unindexed queries from the hot path.

Response time went from 2–3 seconds to 300–500ms.

A SolidWorks knowledge agent grounded on a practising engineer’s own documentation and tutorials, so it answers modelling questions general-purpose models get wrong.

Built as a standing reference for teaching CAD to new engineers, answering from their own material rather than generic advice.

A complete pipeline for e-commerce vendors, not just a model generator. Scan a physical item with a phone, and the platform returns a web ready 3D model, generates plug-in code you can paste straight into a storefront from either the app or the portal, and reports engagement analytics on how shoppers interact with it.

Underneath, the Meshroom photogrammetry pipeline rebuilt into modular RabbitMQ nodes that scale independently, with Draco compression on the output.

Model size cut by 90%, and judged 3rd best Final Year Project at SEECS Open House 2025.