I build the data layer for the agent era.

I’m Andrew Fribush, an engineer who likes the parts of the stack other people find tedious: messy government datasets, fragile ingestion pipelines, and the unglamorous work of making information actually queryable.

Right now I’m building a suite of public-data APIs that turn scattered US federal sources, campaign finance, workplace safety, drug approvals, consumer complaints, patents, the Federal Register, and dozens more, into clean MCP servers and REST APIs that LLMs and humans can both query. Underneath them sits a shared corpus framework I built so a new dataset becomes a production service in days, not months.

The other half of my work is at Abstract, where I build AI infrastructure with an engineering team, speech and video analysis, knowledge systems, customer intelligence, and compliance, for clients like ALL-SAFE, FINCA International, and Saatva. It started as a Shopify Plus shop, and still ships commerce for beauty and DTC brands including Pat McGrath Labs, Credo Beauty, and AUrate.

Before all that I spent a decade founding companies and shipping software for banks, civic campaigns, and institutions. The throughline: I build systems and write about what I learn when they break.