Featured work

Internal data platform supporting the New Mexico Bureau of Geology and Mineral Resources — helping staff ingest, manage, search, and publish water-related datasets through a modern admin UI backed by a FastAPI service layer.

• Refine.dev admin interface for dataset and metadata management
• TypeScript/React frontend with role-based workflows
• FastAPI backend integrating with PostgreSQL/PostGIS data stores

Dagster-based data pipeline that ingests, transforms, and loads water monitoring data from disparate sources into a unified schema — making environmental datasets more interoperable and ready for downstream tools like Ocotillo.

• Python ETL assets with Dagster orchestration and observability
• GCP-based deployment for scheduled and event-driven ingestion
• Designed for reproducibility, testing, and long-running public data workflows

Capstone project building a grounded retrieval system that pairs knowledge graphs, machine learning, and LLMs to help people understand, compare, and evaluate complex legislative information — with explicit evaluation workflows for trust and quality.

• Neo4j knowledge graph with GraphRAG retrieval pipelines
• LLM evaluation framework for measuring explanation quality
• React Native client and FastAPI backend deployed on AWS