Jobs at 120ft Factory AB

Investigate whether operational robot rollout data can improve Vision-Language-Action policy learning for real assembly tasks. Compare operational and teleoperation datasets, analyze success and failure patterns, test fine-tuning with outcome-based filtering or weighting, and assess label reliability. The project is offline and experimental, using existing datasets and training infrastructure, with potential access to a Unitree G1-D for additional rollouts.

Master’s thesis focused on robotic screw driving for a humanoid robot. Compare handheld, dedicated end-effector, and arm- or wrist-mounted screwdriver concepts; select and justify a solution; design and build a powered prototype with microcontroller control; and validate repeated operation through teleoperation using a Unitree G1-D.

Guide robot arms through assembly tasks using teleoperation equipment, repeat tasks to capture consistent training demonstrations, prepare parts and reset workstations, review recordings, flag issues, and collaborate with robotics engineers to improve demonstrations. The role is part-time, paid hourly, and on-site at the Stockholm Factory. No previous robotics experience is required; training and support are provided.

Design and maintain digital twins and simulation environments for robot learning. Build Python and C++ infrastructure for reinforcement learning, imitation learning, synthetic data generation, and sim-to-real transfer. Integrate learned policies, LLM-driven planning, world models, locomotion, dexterous manipulation, and deterministic safety supervisors across multi-robot factory systems.

Architect and implement a deterministic, high-performance Factory Engine for autonomous manufacturing. Build simulation loops, logistics economies, dependency graphs, pathfinding, deadlock resolution, and continuous 3D spatial models capable of handling millions of entities. Apply data-oriented design, entity component systems, cache optimization, and stable tick-rate techniques while integrating simulations with VLA AI models and physical hardware.

Build the observability layer connecting industrial robots, sensors, and AI systems. Responsibilities include capturing telemetry, video, and 2D/3D vision data; developing safe-state projections through sensor fusion; integrating heterogeneous robotic hardware and PLCs; deploying object detection and segmentation models; and creating reinforcement-learning data flywheels for closed-loop industrial automation.

Own and automate the CAD-to-SimReady pipeline, converting manufacturing CAD into optimized OpenUSD assets for physics simulation. Develop collision-mesh simplification, AI-driven material and physics assignment, articulation and joint inference, and solver validation for Isaac Sim. Interface with PhysX and Newton, collaborate on agentic AI workflows, and enable procedural factory-layout generation through the Omniverse Python API.

Own the Skills Supervisor Orchestration layer for a robotics and agentic AI platform. Define system contracts, task sequencing, behavior trees, safety-state projections, operational envelopes, and capability admission gates. Build deterministic supervisory and fault-recovery systems, collaborate with VLA and simulation engineers, orchestrate heterogeneous hardware, and maintain shopfloor dashboards and DfM workflows connecting MES/ERP abstractions to robotics execution.

Generalist robotics engineer responsible for humanoid robot bring-up, calibration, maintenance, teleoperation and data-collection pipelines, simulation integration, rapid prototyping, and hardware testing. The role also serves as Sprint Lead, managing two-week sprints, removing blockers, coordinating scope, and leading physical-hardware demonstrations. Candidates need breadth across mechanical, electronics, firmware, and robotics systems, plus strong Python and familiarity with ROS 2 and modern simulators.