Data Engineer

Context & Vision

In 2026, writing code is no longer the primary bottleneck; managing its complexity and ensuring its reliability is. We are building a highly resilient advertising platform with a very lean, senior internal team, and we intend to keep it that way.

To achieve scale without the overhead of a large engineering department, we rely on an AI-first development paradigm and a philosophy borrowed from the best large-scale open-source projects. Our code is not public, but our governance is theirs: asynchronous communication, exhaustive written documentation, explicit rules, and uncompromising quality gates. It is the only way a small team of humans and AI agents ships serious systems without accumulating technical debt.

This role owns the data engine underneath the platform: the streams, segments and audiences that make advertising intelligent.

The Role

You will own the data and DMP layer end to end — the event streams flowing in, the segment and audience computation, and the analytical stores that feed targeting and reporting. What we hire for is judgment: the problems here are hard in the interesting way, and the interesting part is the tradeoffs you choose.

1. Hard problems, at real-time scale.Computing audiences and measurement over very large, high-velocity event streams, in near-real-time, where being right is not always the same as being exact and every choice has a real cost. We describe the problems; we expect you to bring the approach.

2. Stream-First Pipeline.Own the ingestion and processing pipeline. It is stream-first by design; you keep it that way rather than falling back to batch.

3. Segments & Audiences.Own how raw signals become targetable audiences — how they are computed, stored, refreshed, and exposed to the serving path. This is the intelligence layer of the product.

4. AI-First Delivery & Governance. We expect AI to write most of the code; you are the editor, not the typist. You specify work as explicit, self-sufficient issues, direct agentic tooling to implement, and gatekeep every diff. Schemas, data contracts and Architecture Decision Records are documented. If it is not written down, it does not exist.

 The Tech Environment

- Data: event streaming and stream processing, relational and analytical stores.
- Language: the codebase is a compiled systems language; we care more about how you reason about data than about the language you arrive with.
- Domain: DMP, segments, audiences, identity, measurement.
- Cloud: Google Cloud Platform, Kubernetes.

Profile & Requirements

We are looking for a data-minded engineer who frames a problem before reaching for a tool, owns correctness, and thrives in a lean, high-quality environment. This role is not suited to someone who wants a narrow, hand-fed scope inside a large team.