About the job
You'll work across the whole backend: the SvelteKit app and its Postgres, Redis, and ClickHouse data layer, the Python services that run our ML inference, and the Kubernetes clusters that tie it together across multiple cloud and GPU providers. You'll own these systems end-to-end, from design through to running them in production.
Responsibilities
- Durable execution for long-running workflows. Building the state machines, idempotency, and retry semantics that keep multi-step inference jobs correct when they take minutes to finish and anything can fail mid-flight
- Improving our public API. Designing a clean, well-documented API surface with OpenAPI docs auto-generated from our Zod schemas
- Enterprise product features. ACLs, audit logs, entitlements, and all the "boring" features that organizations with a thousand users expect
Qualifications
Minimum
- You've owned real production systems end-to-end: design, monitoring, metrics, incidents.
- You're comfortable across the stack and unfazed by unfamiliar code, and you have opinions about how to architect things that scale
- You write and talk clearly, and you're happy working without a lot of PM hand-holding
Preferred
- Have GPU or infra experience
- Have run Kubernetes in production
- Come from a creative background