Simulation Engineer
Confirmed on the employer's own hiring board on Nov 11, 2026. First seen by Alion on Sep 24, 2026.
Snapshot
The mission of 120ft Factory is to start the new western industrial revolution. We are building the essential infrastructure to close the "Physical AI Deployment Gap." To train physical AI in simulation, the simulation must be physically accurate. Our core thesis is that the manual CAD-to-Sim pipeline is a massive bottleneck, and we must automate the translation of engineering intent into physics-accurate-enough, machine-readable artifacts.
Description
As the Simulation Pipeline Engineer, you own the translation of the physical world into our digital twin. Raw manufacturing CAD is useless for physics simulations: it is too heavy, lacks material properties, and simulating actual screw threads will instantly crash a rigid-body solver. You are the bridge between elite game-industry asset pipelines and hardcore robot physics. Your mission is to automate the text-to-CAD-to-SimReady pipeline: ingesting raw assets, using agents/VLMs/LLMs to assign physical properties, building solver-friendly collision meshes, and setting up kinematic articulations so our Vision-Language-Action (VLA) models can train on them in Isaac Sim.
Responsibilities
Own and automate the CAD-to-SimReady pipeline, converting raw manufacturing CAD (STEP, etc.) into optimized, physics-ready OpenUSD (USDA/USDC) files.
Develop heuristics and algorithms to automatically generate simplified, solver-friendly collision meshes (e.g., abstracting screw threads and micro-chamfers to prevent physics engine explosions).
Implement VLM/LLM-driven material and physics assignment pipelines (rigid body, friction, restitution, density), building upon frameworks like NVIDIA's content-agents.
Solve the articulation gap: automatically inferring and assigning joints, drives, and kinematic structures (UsdPhysics.ArticulationRootAPI) for robots and moving machinery.
Interface directly with underlying physics solvers (e.g., Newton, PhysX in Isaac Sim) to debug contact dynamics and ensure runtime validation of your generated assets.
Collaborate with the AI orchestration team to enable 'compilable engineering,' allowing LLMs/agents to procedurally generate and modify factory layouts via the Omniverse Python API.
Requirements
Experience in Technical Art, Physics Programming, or Simulation Engineering from the gaming, VFX, or robotics industries.
Deep understanding of 3D asset pipelines, geometry optimization, and how physics engines (PhysX, Havok, Newton, etc.) actually resolve contacts and constraints.
Expertise in OpenUSD (Universal Scene Description) and Python.
A deep, first-principles understanding of real-world physics and kinematics, combined with the pragmatic knowledge of how to fake it efficiently for a solver.
Preferred Qualifications
You have the specific battle scars to know exactly why simulating a physical fastener's threading in a robotic assembly simulation is a terrible idea.
Experience with NVIDIA Omniverse, Isaac Sim, and the UsdPhysics schema.
Knowledge of how to build LLM/VLM agentic workflows for asset generation, treating AI as an interface to compile engineering artifacts.
Experience generating semantic labels for perception AI, or defining non-visual material properties (lidar/radar).
You view the role of an engineer as shifting from a manual 'tool operator' to a 'system architect' who builds pipelines.
Compensation
We offer a top-tier Stockholm salary combined with an early-stage equity grant as part of the founding engineering team. Comprehensive relocation support and EU Blue Card sponsorship provided.