Open position
Serving stack
Build the batching, scheduling and caching that keep cost and latency predictable under real traffic, for every model in the catalog.
What you'll do
- Run the job queue, batching and scheduling behind the API's asynchronous jobs.
- Keep latency and cost predictable when traffic spikes or a model changes.
- Build the storage, caching and webhook delivery customers rely on.
- Instrument the stack so problems show up before customers notice them.
What you bring
- Experience building and operating backend systems under real load.
- Comfort with queues, containers and GPU scheduling.
- Clear thinking about failure: retries, timeouts and what happens when a machine dies.
Good to have
- Experience serving machine-learning models in production.
- Work on autoscaling GPU fleets.