Robotic Research Engineer

About Us

Trener Robotics is building the software stack that lets industrial robots become easier to deploy, operate, and improve over time. T-Labs is our robot-learning team, focused on training and integrating Vision-Language-Action models and learned manipulation policies for real industrial tasks. The team is split between our headquarters in San Jose, California and our lab in Trondheim, Norway.

Our goal is general-purpose manipulation for industrial robots, starting with machine-tending part handling: collecting high-quality robot data across a growing range of tasks, training VLA policies from strong open backbones, and deploying learned skills into Acteris, our edge runtime for robot operation.

About the Role

We are looking for a Robotics Research Engineer to own the physical learning loop in our San Jose lab. You will keep the robot systems, demos, teleop rigs, UMI data collection setups, cameras, grippers, and sensors running so the team can collect useful data and evaluate learned robot policies on real hardware.

This is not a pure lab manager role and not a pure technician role. You should be hands-on enough to wire, calibrate, fixture, debug, and operate robot cells, but technical enough to work closely with robot learning engineers, systems engineers, and controls engineers. Your job is to make sure experiments happen, demos stay ready, and the data that reaches the training pipeline is usable.

This is also not a pure ML job, but ML experience is required. You should be able to start training runs, run policy evaluations on the collected data and on the robot, and read the results well enough to tell the team whether the data or the policy is the problem.

You will be the on-site owner of the San Jose robot cells and will coordinate closely with the Trondheim lab so that both sites run the same collection protocols, calibration procedures, and data formats.

How You'll Move the Mission Forward

• Own day-to-day readiness of robot learning lab systems: robot arms, grippers, cameras, trackers, force/torque sensors, safety equipment, and compute boxes.

• Keep demos and data collection rigs operational, calibrated, documented, and ready for scheduled runs, including customer, investor, and partner demos at the San Jose office.

• Run and support teleop data collection using VR headsets, trackers, robot controllers, and recording tools.

• Support the UMI data collection by testing hardware, calibration, synchronization, ergonomics, and data quality in real collection sessions.

• Collect high-quality robot demonstrations and autonomous execution logs across a broad and growing range of manipulation tasks, from machine-tending part handling to general pick, place, insertion, and reorientation.

• Perform first-pass data QA: verify timestamps, camera views, robot state, action logs, task metadata, labels, and discard reasons before data enters the training dataset.

• Work with the data platform team to make sure collected episodes upload correctly and carry the metadata needed for training, evaluation, and model lineage.

• Work with robot learning engineers on what data is useful for VLA training, policy evaluation, recovery cases, and failure analysis.

• Work with systems and controls engineers to debug policy execution, behavior tree integration, robot motion, safety stops, and runtime logging.

• Design, fabricate, and maintain fixtures, mounts, adapters, test objects, and cell layouts for repeatable experiments.

• Document setup procedures, calibration steps, collection protocols, failure modes, and lab operating practices, and keep them in sync with the Trondheim lab.

What You Need to Succeed

• Hands-on experience building, operating, or maintaining robotic manipulation systems in a lab, company, or advanced academic setting.

• Strong practical understanding of robot arms, grippers, cameras, sensors, calibration, and safety around physical robots.

• Ability to debug real robot systems across hardware, software, networking, sensors, and operator workflow.

• Programming ability in Python; enough C++ or ROS 2 familiarity to work effectively with robotics software engineers.

• Comfort with data collection for robot learning: demonstrations, episodes, timestamps, camera streams, robot state, actions, labels, and quality checks.

• Working ML experience: able to launch training runs from an existing pipeline, run evaluations, and interpret basic training and eval metrics.

• Practical mechanical/electrical skills: wiring, mounting, 3D printing, fixture setup, connector troubleshooting, and basic CAD.

• Clear documentation habits and the discipline to keep lab setups repeatable.

• Business fluency in English.

• Authorization to work in the United States.

What Will Differentiate You

• Experience collecting data for imitation learning, diffusion policies, VLA policies, ACT, or similar robot learning methods.

• Experience with teleop systems, VR controllers, motion trackers, UMI-style collection rigs, or learning-from-demonstration workflows.

• Hands-on experience with industrial or collaborative robots such as Universal Robots, ABB, FANUC, KUKA, or similar.

• Experience with camera calibration, multi-camera recording, depth cameras, wrist cameras, or vision-system debugging.

• Familiarity with PyTorch, LeRobot-style datasets, RLDS, HDF5, Parquet, or other robot-learning data formats.

• Experience deploying or evaluating learned policies on real manipulators.

• Experience with industrial communication protocols such as Ethernet/IP, Modbus, OPC-UA, or vendor robot APIs.

• Experience with CAD tools such as SolidWorks, Fusion 360, or Onshape, plus 3D printing or basic machining.

• Experience working with a distributed team across time zones.

Why Join Us

You will own the lab systems that make robot learning real. When the team needs better data, a demo-ready cell, a calibrated teleop setup, or a policy tested on hardware, you are the person who makes it happen.

This role sits at the center of T-Labs' next six months: manipulation data collection, VLA model training, UMI validation, data pipeline flow, and Acteris integration. Your work will directly determine how fast the team can turn robot data into deployable learned skills.