The role
- COMP
- $200K - $280K
- EQUITY
- Competitive Equity
- LOCATION
- New York · Seattle
- WORKPLACE
- Hybrid
- EXPERIENCE
- 3+ years
- VISA
- None, Visa transfers
- STACK
- Kubernetes, Kafka, SQS, LLM infrastructure
- INDUSTRY
- Data, AI
/roles — ROLE_5
Member of Technical Staff
High-fidelity trajectories and environments for frontier AI research
The company
Frontier data laboratory building deterministic RL environments and pixel-perfect trajectory datasets for training computer-use agents.
- STAGE
- early-stage
- TEAM
- founding team of <10
- FOUNDED
- 2025
JD — the work
About the role
Infrastructure at the frontier of agent evaluation: orchestrate hundreds of simultaneous agent runs — each with its own stateful environment — at a combined throughput of ~100M tokens per minute. You'd own the dispatch layer and help design the environments and tasks that genuinely stress-test frontier agents.
What you'll do
- Own agent orchestration at scale: queueing, concurrency control, failure handling across the fleet
- Design realistic environments and evaluation tasks that push agents to their limits
- Build new evaluation modalities and integrations with external orchestration frameworks
- Stay on the leading edge of a fast-moving agent-evaluation space
What they're looking for
- Distributed-systems experience in production
- Container orchestration (Kubernetes) and message-driven architectures (SQS, Kafka)
- A track record of running LLM workloads at scale