title: pol_watch status: active oneLiner: An open-core data layer for tracing money through American campaign finance, federal and state. needs: [backend, policy]
What we're building
Campaign-finance disclosure in the United States is fragmented across the FEC and fifty different state systems. pol_watch is the canonical data infrastructure underneath that mess — ingestion, schema normalization, legal-entity resolution, and a read-only API — so journalists, watchdogs, and downstream products can plug in without having to re-solve the plumbing problem.
Where we are
The federal FEC spine is shipping behind a FastAPI gateway. The first state — Washington — is in flight, on Splink-based probabilistic entity resolution. We deliberately started with Washington PDC because it's the best-documented state CF portal in the country; proving the pipeline end-to-end here is the gate before we tackle California, Texas, and New York.
The ingestion and resolution code is Apache-2.0. Investigative tooling that sits on top is in a separate, source-available repository.
What you'd do
If you join as a backend contributor, the critical path is the entity resolution layer — every additional state we ingest depends on it. If you're a policy or domain contributor, what we need is help mapping jurisdiction-specific quirks: the definition of "contribution," what counts as an in-kind, when amendments get filed, and which fields a given state portal silently drops.