IRS 990 Pipeline
Automated pipeline for processing IRS Form 990 nonprofit financial filings at scale using Cloudflare Workers, D1, and Queues to process millions of records.
The IRS publishes Form 990 data for all US nonprofits, representing millions of records going back decades. This pipeline ingests, normalizes, and indexes that data to make it queryable and useful for researchers.
The pipeline runs entirely on Cloudflare infrastructure. Workers handle ingestion and initial parsing. Queues buffer work during high-throughput batch runs. D1 stores the normalized records with indexes optimized for the common research queries: find all orgs in a geographic area, find all orgs with a given officer, find financial trends for a specific entity.
Processing includes cross-referencing EIN numbers across years to build longitudinal financial profiles of organizations.