Data Engineer

The role

Everything we do runs on one asset: a database of every company and investor out there, every funding round, the news that matters, and how they all connect - plus the raw context underneath: every email and call transcript, linked to the right people and companies. It is a knowledge graph and a memory in one. Our matching, our outreach, and our agents are built on top of it, and it grows faster than anyone can own it on the side. You become its owner. You design it, scale it, keep it clean, and turn it into the single source of truth that everything reads from. To be clear about the shape of this seat: it is not a reporting or analytics warehouse. It is the memory a live product thinks with, built for one reader above all: agents retrieving exactly the right fact at the right moment. One honest filter before you apply: if the database you are proudest of tracked shipments, sensors, factory lines, or compliance - however well you built it - that is a different seat. If it tracked companies, investors, deals, and the people and conversations around them, keep reading.

What you will own

• The database itself: Postgres and Supabase with hybrid search, schema design, modeling, scaling, and performance as it grows without a ceiling. The agents that read it run on Pydantic AI and the Claude Agent SDK, on AWS. We are consolidating into pgvector, not buying a vector DB.
• Data quality end to end: validation gates for vendor and third-party data, dedup, entity resolution, provenance, monitoring.
• The communications layer: raw emails and call transcripts stored, linked to the right people and companies, and searchable.
• Ingestion and enrichment pipelines: funding rounds, market news, and contact and company research at scale, engineered for cost and freshness.
• The knowledge graph: companies, investors, funding rounds, and news as entities and relationships - node and edge tables in Postgres, provenance on every fact.
• The unified data layer: one clean spine that every campaign, agent, and product feature reads from.

You are a fit if you

• Have owned a database of companies, people, deals, or the communications between them - a CRM source of truth, a market or deal intelligence graph, an enrichment layer - that a live product, agents, or a sales team read from. Serving dashboards is a different job than this one.
• Are strong in SQL and Python, with real pipeline work behind you: ingest, transform, dedup, enrich.
• Have caught bad data before it hurt the business, and can tell us how.
• Think in schemas and contracts, and design for the queries of a year from now.
• Have modeled entities and relationships at scale - companies to investors to rounds to people - and kept the connections queryable as the sources multiplied.