AI Lab Tech Engineer

Role: AI Lab Tech EngineerLocation-Fremont, CA (Local Candidate Onsite/Hybrid work)
Job Type-Long Term Contract

Were looking for an Infrastructure Engineer to own the execution layer beneath our RL environments: the systems that let an agent operate inside a realistic, multi-tool world coherently for hours or days.

This is a hard systems problem disguised as an AI job.
As the tasks agents can complete keep lengthening, the environments that train them have to stay coherent across far longer horizons than anything that exists today.
That means sandboxing and isolation you can trust, execution thats fast and cheap enough to run at training scale, and the ability to snapshot, restore, inspect, and branch a running environment instead of treating every rollout as one-shot. Youll build the platform that makes all of this possible.

Youll work closely with our research and data teams, and directly with frontier labs and enterprise customers, to turn environment designs into infrastructure that runs reliably in production.

What Youll Do

1. Environment Execution & Sandboxing:

• Design and own the sandboxing and execution layer that environments run inside. Build systems to snapshot and restore environment state (disk, process, and where relevant memory and accelerator state) so runs can be paused, resumed, inspected, and branched rather than executed once.
• Develop the machinery to detect failure modes early in a rollout (reward hacks, infra faults, fairness issues) and to revert to a known-good state, patch, and continue.
• Extend execution to long-horizon and multi-node environments, where an agent operates across many tools and services over hours or days.

2. Performance & Scale

• Own the performance characteristics of the platform: throughput, latency, and cost-per-rollout at scale.
• Drive utilization and scheduling so we can run far more environment rollouts per dollar without sacrificing reliability.
• Profile and remove bottlenecks across the stack, from container startup to environment teardown.
• Build the observability that lets us understand whats happening inside thousands of concurrent, long-running rollouts.

3. Environment Platform

• Build and maintain the framework for specifying, packaging, and deploying RL environments which is used by both humans and agents authoring environments internally.
• Create the tooling that lets researchers and environment authors debug a specific failure across hundreds of long agent traces.
• Deploy large / small models on on-prem hardware

4. Collaboration & Production Excellence

• Scale prototypes into production systems with reproducible workflows and high engineering standards.
• Write the documentation and tools that let internal teams and external users build on the platform.

What Were Looking For

1. Systems & Infrastructure

   Looking to get Placed? Try our Placement Guarantee Plan

   Get Hired

• Strong track record building production systems or research infrastructure at scale: distributed systems, execution engines, container/sandboxing infrastructure, or similar.
• Deep comfort with the systems layer: containers and isolation (e.g. namespaces, cgroups, VMs, gVisor/Firecracker-style sandboxing), filesystems, process and state management.
• Experience making systems fast and cheap profiling, scheduling, resource utilization, and cost optimization at scale.
• Proficiency with cloud platforms (Google Cloud Platform, AWS) and distributed computing.
• Strong engineering fundamentals and a systematic approach to testing, validation, and reliability.
• Experience of deploying large / small models on on-prem hardware

2. Execution & Ownership

• Comfort operating in ambiguity.
• Strong Python skills; comfort in a systems language (Rust, Go, or C++) is a plus.
• Ability to use modern tools such as Claude Code effectively.

3. Collaboration & Communication

• Excellent communication skills for working with research teams and enterprise customers.
• Ability to translate between research needs and infrastructure requirements.
• Comfortable presenting technical work to diverse audiences.