InLoop builds warehouse robots that run fully autonomously in customer warehouses today. The fleet is growing quickly, and the demand is huge.
We're hiring an embedded engineer to own the software that runs closest to our hardware, on the robot and on the device we use to teach it: a UMI style handheld gripper that records what a person does so our robots can learn it. You'll work with our CTO, our hardware team and our ML team.
This is a hands on role. Some weeks you're bringing up a new board, some weeks you're on the bench chasing a bus error, and some weeks you're watching your device collect data in a real warehouse.
**What you'll work on first**
**On the robot**
\- **Our own compute board.** Bring up our Jetson carrier board: BSP, device tree, GMSL2 camera drivers with our camera vendor, CAN and I/O, replacing the separate devices the robot carries today.
\- **Motor and sensor buses.** Make the links to the arms and sensors reliable: drivers, timing, error handling.
\- **Safety state.** Build the software side of the safety system with our electrical engineer, so the robot always knows its state and reports it.
**On the data collection device**
\- **Handheld gripper.** Build the electronics and software inside it: camera, IMU, pose tracking, gripper width sensor and recording.
\- **On device tracking.** Choose and build real time 6 DoF tracking, then prove its accuracy and drift against our robot arm.
\- **Sync and calibration.** One clock for every sensor, a left and right device in sync, and calibration quick enough to run on every unit.
\- **Recording and upload.** Clean episode files our ML team can train on, live feedback when tracking drops, and upload that never loses an episode.
**Across both**
\- **Same eyes on both sides.** The same camera stack on the robot and the device, so the robot sees exactly what the device saw.
**What we're looking for**
\- 3+ years writing firmware or embedded software for real robots or similarly unforgiving hardware
\- Embedded Linux on Jetson or similar boards: drivers, device tree, camera bring up
\- CAN, Ethernet and USB, and the tools to debug them (scope, logic analyzer)
\- Visual inertial odometry or SLAM running live on embedded hardware (ORB-SLAM3, OpenVINS, VINS or similar)
\- Multi sensor time sync and calibration: cameras, IMU, encoders
\- C and C++, plus Python for tooling
\- You can read our schematics and debug hardware on the bench with a scope and logic analyzer
**Nice to have**
\- Bringing up a custom Jetson carrier board
\- UMI or similar handheld data collection devices
\- GMSL2 or MIPI CSI cameras
\- Battery powered devices
\- Robot learning data formats (MCAP, LeRobot or similar)
\- German
**How to apply**
Apply here and include a link or files of something you built that runs on real hardware. That tells us more than a CV.
Visa sponsorship is offered for this role.
