Seed-stage team building simulation environments for frontier AI research.
Member of Technical Staff — Infrastructure & ML Platform
Within your first year, you'll own the orchestration layer that lets researchers spin up thousand-node training clusters before lunch and trace exactly where a tensor went sideways. You'll harden a Kafka pipeline moving terabytes of trajectory data between edge simulators and GPU farms, then shave latency until the bottleneck is physics, not plumbing. The work sits at the intersection of ML systems and site reliability: every millisecond you reclaim is another experiment someone else can run.
What they're looking for
- 3+ years running production Kubernetes at scale — multi-region, multi-tenant, or bare-metal-on-cloud
- Deep comfort with streaming architectures; you've debugged Kafka partition rebalances at 2 a.m. and lived to tell
- Fluency in at least one systems language (Go, Rust, or Python with C extensions) and the metrics to prove it
- Track record of building self-service platforms that let ML engineers ship without filing tickets
- Experience with GPU cluster scheduling or training-workload optimization in a research environment