Abhishek Yadav

Led end-to-end development and production deployment of Nexus, an AI-powered document verification platform for construction submittals, implemented as a client-facing technology solution with 99.5% uptime and processing over 10,000 documents per month.
Designed and implemented a Retrieval-Augmented Generation (RAG) pipeline using Azure Document Intelligence, custom vector embeddings, and dual similarity search to extract structured data from multi-page construction PDFs and technical specifications with 95% accuracy.
Fine-tuned and evaluated multiple open-source large language models on 50,000+ construction documents to improve domain-specific terminology understanding, entity recognition, and compliance checking (89% accuracy).