Built proprietary NLP algorithms to screen Intel's full vendor contract portfolio — 100,000+ contracts — for clause-level risk signals across multiple geographies. Reduced annual risk exposure by $33M and won the 2020 Gartner High-Tech Chainnovator Prize.

Constructed and delivered an automated contract question-answering pipeline built with an LLM — enabling legal and procurement teams to query any contract in plain English with no manual review. Eliminated 500+ audit hours and expanded program coverage by 30%.

Built fuzzy matching algorithms to consolidate millions of parts across supply chain datasets into a supplier catalogue. Powered dashboards enabling supplier consolidation and pricing leverage — $7M in measurable procurement savings.

Released categorization and matching models to production on Azure and Databricks — automating classification of tens of thousands of previously unclassified parts. Eliminated critical data gaps that had caused years of reporting inaccuracy.

Optimized time series forecasting model for spare parts utilization through predictive feature engineering — achieving 10% error reduction across 100,000+ parts for manufacturing and data center infrastructure. Reduced excess inventory, minimized equipment downtime, and enabled cross-region parts-to-machine matching.