The Role
As a founding member of the team, you will build the software systems that generate the data needed to teach our models how the physical world works. You will architect the bridge between virtual environments ranging from specialized physics simulators to broad interactive software and our high-performance training pipelines.
Your work provides the "imagination" for our world models, allowing them to learn from a breadth of physical scenarios that are difficult or impossible to capture in the real world. You are a systems-level generalist who can reach into an engine's internals to extract what we need.
Job Responsibilities
- Area: Design and scale the infrastructure for virtual environments and synthetic data generation.
- Ownership: The software stack that transforms simulated interactions into high-fidelity training telemetry.
- Focus: Integrating diverse interactive engines and building procedural tools to generate a vast array of physically grounded scenarios.
- Operations: Optimizing simulation performance to maximize data throughput and ensuring the technical consistency of virtual datasets across the cluster.
Minimum Qualifications
- Extensive experience in software engineering for simulation, robotics, or complex 3D systems.
- Expert-level experience with 3D runtimes and physics-driven environments (e.g., Unreal, Isaac Sim, or other interactive engines).
- Expert with Python and C++ / Rust.
Preferred Qualifications
- Proof of outstanding contribution to open-source projects in graphics, simulation, or procedural generation.
- Experience reverse-engineering or "modding" interactive software to extract internal state and telemetry for research purposes.
- Familiarity with scaling simulation compute in cloud or distributed environments.