Data Scientist; EO and SAR

Role Overview

The Senior Data Scientist on NGA Maven will apply statistical rigor, geospatial expertise, and software development skills to ensure that EO and SAR data feeding Maven’s models is carefully selected, processed, and partitioned for maximum analytical value.

• Conduct analysis of Maven data holdings to evaluate impact on ML model development: assess data diversity, training/test/validation splits, and recommend partition strategies for effective model performance evaluation.
• Integrate emerging EO/SAR sensors into Maven pipelines — evaluating metadata, format, and schema differences and building the ETL workflows to ingest new data.
• Build and maintain labeling campaign management tools: track unlabeled vs. labeled data, campaign status, and enable API‑based transfer of label task information between platforms.
• Develop and apply imagery curation algorithms and analytic tools for acquisition prioritization and chipping, including web scraping for curation per Maven data priorities.
• Build geospatial visualization and filtering tools that integrate with the existing Data Management Platform.
• Apply ETL and statistical methods to support data cleansing, analyst workflow efficiencies, and intelligence requirement fulfillment.

• Integrate emerging EO/SAR sensors into Maven pipelines; assess ETL process changes needed for new metadata, formats, and data structures.
• Lead the design and application of methods to identify, collect, process, and analyze large volumes of data to build and enhance products, processes, and systems.
• Evaluate and recommend solutions for partitioning emerging sensor data into effective training, test, and validation splits for ML model development.
• Build imagery curation algorithms and web‑scraping tools per Maven data priorities; integrate multiple data and intelligence sources to address gaps.
• Develop labeling campaign management software tracking unlabeled/labeled data status and enabling API‑based movement of label task information between platforms.
• Build tools to filter and visualize data geospatially, enable feedback entry, and integrate with the existing Data Management Platform.
• Conduct analysis of overall Maven data holdings to support development of performant AI/ML models satisfying operational user requirements.
• Evaluate, monitor, and recommend ways to partition training, test, and validation splits for effective model development and performance evaluation.

• Active TS/SCI clearance.
• Minimum 10 experience points required (see Experience Point Requirement below).
• Experience working with AI/ML technologies and data systems.
• Experience working with multiple file types including geospatial file formats, JSON, and XML.
• 3+ years of experience performing quantitative analysis, developing visualizations, and processing complex data to create data‑driven insights; includes data manipulation and ETL experience with SQL and No
  
  SQL.
• Development experience in Python and other languages for data cleaning and manipulation.

• Experience applying NLP algorithms to extract data from documents.
• Experience with NGA analytic modernization efforts: SOM, computer vision, automated collection, or automated reporting; familiarity with data standardization best practices.
• Demonstrated expertise in mathematics, statistics, and quantitative analysis; experience with classification, regression, clustering, data reduction, and causal modeling techniques.

• Python
• SQL / No
  
  SQL
• AI/ML Technologies
• EO/SAR Data
• Geospatial File Formats
• JSON / XML
• ETL Pipelines
• Data Visualization
• NLP (Preferred)
• Computer Vision (Preferred)
• Statistical Modeling (Preferred)

• You think statistically about data — you understand why training/test/validation splits matter and you design them with model performance in mind.
• You build software that other people can use — your tools are reliable, documented, and integrate cleanly with existing platform infrastructure.
• You are comfortable with ambiguity around emerging sensors — you can assess an incomplete spec, make sound assumptions, and build toward…