At Public Storage, I architected a hybrid LLM gateway strategy that routes requests to self-hosted open-weight models or proprietary frontier APIs, reducing projected inference spend while maintaining output quality.
I deployed a multi-cloud GraphRAG platform on AWS and GCP, combining vector databases, knowledge graphs, and hybrid search to improve context recall. I also built model distillation and PEFT/QLoRA pipelines to align smaller open-weight models for operational and product telemetry.
At Datacog, I defined a multi-cloud data strategy and deployed a cloud-native analytics platform using BigQuery and serverless transformation pipelines. I developed ML pipelines for predictive analytics and forecasting, and integrated IoT streams for operational insights.
Earlier, at MSC Industrial Supply, I improved production pipeline quality and supported data migrations during two acquisitions. At FamilySearch, I developed AWS data pipelines and improved query performance through Oracle PL/SQL and warehouse optimization.

