Methodology

How we build and maintain the dataset — and how you can cite it.

Sources

AI/ML keyword filter

Contracts are included when the award description matches one of these terms (case-insensitive substring): artificial intelligence, machine learning, deep learning, large language model, generative AI, natural language processing, computer vision, predictive analytics, neural network, MLOps, AI, ML, LLM.

The filter is intentionally broad to capture both pure-play AI work and AI-enabled systems (e.g. predictive maintenance, autonomous platforms). False positives are minimized by also requiring NAICS codes in the IT/R&D family.

Date range and amounts

The scraper asks USAspending for awards from Oct 1, 2023 (fiscal year 2024) to today. USAspending also returns older awards that had contract activity in that period, so the earliest start date in the data is Apr 5, 2018. Each amount is the award total that USAspending reports, not only the money spent after Oct 1, 2023.

Refresh cadence

The pipeline runs every 6 hours on a GitHub Actions cron, and each data commit triggers a rebuild. A failed run leaves the previous data in place. Contract data on this build was last updated Oct 10, 2026. The SAM.gov opportunities feed has been paused since May 2, 2026.

License

Underlying federal data is public domain. Our compiled, normalized dataset is released under CC0 1.0 Universal — no attribution required, though appreciated.

How to cite

For academic and journalistic use, please cite as:

Hegde, N. (2026). GovAI Contracts: Federal AI/ML Award Dataset.
https://govai-contracts.nandanhegde1096.workers.dev
Last updated: 2026-10-10.
Records: 756.

Bulk download

Reporting issues

Found a bug, missing data, or wrong classification? Open a GitHub issue.

Independence

This is an independent project. Not affiliated with the U.S. government, USAspending.gov, SAM.gov, or any vendor listed.