Back to experience

NASA IMPACT
Research Engineer (ML & Backend) | Part-Time · Huntsville, AL
Aug 2024 — Present
NASA team prototyping advanced technologies to expand the use of Earth observation data.
www.earthdata.nasa.gov/about/impactAbout
IMPACT (Interagency Implementation and Advanced Concepts Team) is part of NASA's Earth Science Data Systems Program, based at Marshall Space Flight Center in Huntsville, Alabama. It prototypes new technologies — including machine learning and AI — to maximize the scientific return of NASA's Earth observation data and improve data discovery and stewardship. Its work spans foundation models, STAC catalogs and data-management best practices.
Notable: Collaborated with IBM Research on geospatial AI foundation models.
What I worked on
Project: Data Acquisition Request System- Drove a 30% increase in data delivery speed to clients through the implementation of the new Tasking Management System, significantly improving data ingestion and request handling for the NASA CSDA program.
- Initiated and built an agent-based data search project leveraging LLMs with Retrieval-Augmented Generation (RAG) using LangChain and LangGraph, enabling users to intelligently search and retrieve specific data from a massive STAC metadata catalog.
- Developed a RAG-based automatic metadata file generator for the CSDA program, utilizing local LLM models with Weaviate and ChromaDB as vector databases and maintaining a knowledge graph for NASA Spatiotemporal Asset Catalog compliance.
- Managed scalable infrastructure across AWS (S3, EC2, RDS, Lambda, Step Functions) to ensure a robust and highly available system using technologies including Next.js and FastAPI.
- Implemented data ingestion pipelines for new vendors, including Satellogic and Maxar DEM, within the CSDA system.
- Developed and pioneered the new NASA CSDA tasking management system, utilizing a Django/Airflow backend and a React and TypeScript frontend to manage commercial satellite data requests and ingestion.