Founding Engineer, Commerce Data Supply Chain

Ingestion from sources you do not control

The system must ingest data from multiple, uncontrolled sources, each with its own schema, lag, and error patterns.

Entity resolution

The system must resolve entities such as merchants and codes that arrive under different names or with conflicting details.

Freshness as an economic frontier

The system must manage the decay of commerce facts over time, ensuring that only fresh records are used.

Agent validation fleet

• Design and implement LLM pipelines for row-level validation.
• Write evaluations to ensure the fleet's precision.
• Price the fleet based on its operational costs and benefits.

Walking in, you must already own

• Large-scale ingestion.
• Entity resolution and deduplication.
• Data quality Service Level Agreements (SLAs).
• Pipeline unit economics.

Growth opportunities

• Grow into agent fleets as a production workforce.
• Develop LLM evaluation design.
• Run a data platform like a P&L.