GovDataHub — A Commercially Reusable Census Data Service

Government data is public. Finding it in a form you can actually use is another story.

I built GovDataHub to put a small, stable interface in front of U.S. Census datasets: a catalog for discovery, a normalized query contract, cached results, and source context that stays attached to every response.

The implementation is currently a private working prototype. The repository is intentionally not public, but the product boundary is straightforward enough to describe: make public data easier to find, query, and reuse without pretending that an aggregator owns the underlying source.

The problem

When I research a topic — housing, transit, agriculture, whatever — I end up with:

• Browser tabs across several .gov dataset portals
• Variable names and geography syntax that differ by source
• CSV downloads with little context about vintage, retrieval time, or provenance

GovDataHub gives those requests a common workspace. A user can choose a dataset, discover variables, shape a Census query, inspect the returned table, and keep the source URL and cache details with the result.

What is implemented

The first service slice includes:

• A catalog covering ACS, business, and other configured Census datasets
• Variable discovery with labels, concepts, predicate types, and source links
• Normalized query endpoints for get, for, in, time, and additional predicates
• Cache-aware responses with fetch time, expiry, checksum, and row count
• Saved queries and repeatable catalog and refresh jobs
• An accessible responsive explorer with loading, error, empty, JSON, and CSV states

The stack is intentionally plain: FastAPI, async SQLAlchemy, Alembic, HTTPX, Pydantic settings, and a dependency-free browser UI. The boring pieces are useful here because the data model and source boundaries matter more than a large frontend framework.

Commercial reuse

The initial sources are U.S. Census statistical datasets, which are generally available for commercial reuse. GovDataHub is designed to preserve the responsibilities that come with that reuse:

• Keep the U.S. Census Bureau attribution and source links
• Follow API terms, rate limits, and dataset-specific notices
• Avoid implying agency endorsement
• Keep provenance visible when a result is copied into another product

The service adds organization, queryability, and operational context. It does not add a non-commercial restriction to the underlying Census output.

What remains

The prototype is not a public hosted API yet. Before operating one, I would want to choose a narrow audience, add explicit access controls and quota management, define a refresh budget, and measure whether a stable query facade is more useful than direct Census access for that audience.

GovDataHub sits alongside other side projects — SeedStarter for gardening, Browser Listener for local session capture, and Wiggum for auditable software iteration. Each solves a narrow problem I have repeatedly encountered.

Read the project case study for the current service boundary and design signals, or start with the Census dataset documentation.