Patent Analysis Agent
RAG-based patent automation for enterprise
@ Microsoft
Overview
Patent teams at enterprise clients spend hours reading and cross-referencing long patent documents. As a high-priority cloud-native PoC for SK Chemicals, I architected a RAG-based AI agent that ingests patent PDFs, retrieves the relevant claims and prior art, and produces structured summaries - built to handle enterprise-scale document volume on Azure.
Gallery
Highlights
RAG over patent corpora
Retrieval-augmented generation grounds every summary in the source patent text, keeping citations traceable for legal review.
Serverless document pipeline
An Azure Functions pipeline splits large patent PDFs and re-uploads processed sections to SharePoint via Microsoft Graph - keeping ingestion scalable and hands-free.
Enterprise PoC, real stakeholders
Scoped, built, and demoed for SK Chemicals as a lead enterprise engagement during the Microsoft internship - from requirements to a working agent.