About the Company
The client is an applied AI company in San Francisco building the training-data infrastructure behind one of the fastest-moving areas of AI. Their customers are most of the large AI labs and several of the biggest technology companies, and the work sits unusually close to the frontier: the datasets this team produces set the ceiling on what those customers’ models can do.
Founded in 2024 by a technical team out of one of the best-known names in AI data, they are now around fifty people, closed a Series B with tier-one funds last year, and passed an eight-figure revenue run rate inside their second year. Engineering is small and senior and reports into a founder. This role exists because product surface and data volume are both growing faster than the team: they need a hands-on lead to own the platform and raise the standard as the team scales.
The Role
Own the platform that turns terabytes of raw data into the datasets frontier AI models are trained on.
This is a hands-on tech lead seat in product engineering, not a distant architecture role. You ship to production daily, you sit next to the researchers and operators who use what you build, and the technical standards the next ten engineers inherit are yours to set.
You own the product platform end to end: the pipelines, the customer-facing tooling on top of them, and the reliability of both. The seat reports to a founder, and it is one of a small number of senior engineering hires planned this year, so the scope is real from day one.
What is actually unsolved
- Pipelines that comfortably move terabytes today have to survive the next order of magnitude without cost scaling with them. That architecture has not been drawn yet.
- Customers hold enormous unstructured datasets and can barely see inside them. Search, quality signals and insight at that volume is an open product problem, not a feature backlog.
- Quality control still leans on expert human review. An automated evaluation layer, mixing LLM-based and signal-level checks that customers actually trust, does not exist yet.
- The team is growing past the point where standards can live in the founders’ heads. Someone has to write them into the codebase and the culture. That someone is this hire.
Responsibilities
- Lead feature development across the stack and ship to production daily.
- Design, build and scale the systems that process terabyte-scale datasets and turn them into customer-facing insight.
- Build and evaluate tooling that automates quality checks, combining LLM-based and signal-level approaches.
- Drive the iteration loop with research and operations, from data-collection interfaces through to delivery.
- Mentor engineers and set the technical standards the team will carry as it doubles.
Requirements
- Six or more years building product-focused backend or platform systems, with recent years at senior level.
- Has designed and run production systems processing terabyte-scale data, and can walk through the numbers: volumes, throughput, cost, what broke.
- Has taken at least one product from zero to paying customers and can say what it earned.
- Has worked at startup pace inside a reputable high-growth company. Ships fast and iterates rather than polishing in private.
- Hands-on and current in a modern TypeScript stack in production: Node, a React-family framework, Postgres, a major cloud.
- Computer science degree from a leading US program. A client requirement, applied strictly.
- Able to work from the San Francisco office full-time, or to relocate on a clear timeline.
Nice to Have
- Signal processing or speech background, academic or professional.
- Has shipped and operated ML models in production.
- Has led or managed engineers in a high-growth environment.
Culture
Small, senior and tight-knit, in one San Francisco office every day. Production ships daily and the loop between building something and watching a customer use it is short. The founders are technical and close to the work, which cuts both ways: there is no layer to hide behind, and no platform-only lane where someone else worries about the product. People who want remote flexibility or big-company cadence will not enjoy this. People who want their fingerprints on the standards a company scales with generally do.
Equity
Yes. Stated as competitive; the exact band is confirmed early in the process
Location
San Francisco, CA
Working model
On-site, office-based, full-time
Sponsorship
US work authorization required. Transfers supported, including H-1B transfer and OPT