Senior Robotics Engineer, Dexterous Manipulation
Analog
- Abu Dhabi, United Arab Emirates
- On-site
- Senior Level
- 5+ years
- Posted
Skills
Required
- grasp planning
- robot manipulation
- 6D
- object pose estimation
- kinematics
- MoveIt
- C++
- Python
Preferred
- imitation learning
- VLA policies
- Humanoid / mobile manipulation
Why Analog, and why now
Analog builds robots that work in the real world, deployed in Abu Dhabi. Our robots are already out in clinics and outdoors. They are getting good at moving through the world. The next step is getting them to do things with their hands.
The robot comes with dexterous, multi-finger hands already built, plus a set of cameras and touch sensors in the wrists to guide grasping. What's missing is the person who makes those hands actually useful: picking things up, handling them, and recovering when it goes wrong. If you've shipped that on real hardware, this is a rare chance to own it from day one at a 77-person company, in one city, with a team that's here because they want to be. Ana, our neural agent, is also meant to get a robotic body someday, and manipulation is a big part of how she'll get there.
About Analog
- We want depth in one hard problem, not a bit of everything. That's what takes us from impressive demos to capabilities that win new customers.
- You're the manipulation specialist.
- You'll shape how we approach grasping and manipulation, and your decisions will stick.
- Real hardware, real conditions.
- You test on actual robots in the same building, then watch them go into the field.
- Flat structure.
- No layers between you and the people making decisions, so good ideas move fast.
- You build the intelligence on top of existing dexterous hands instead of designing hardware from scratch.
Responsibilities
- You'll build the full manipulation pipeline on top of our existing dexterous hands:
- Grasp planning and selection
Arm and hand trajectory execution
- Failure detection and recovery, using force and tactile feedback
- You'll connect all of it to the hand drivers and motion control stack, then get it running reliably on deployed robots, in real conditions, not just in simulation.
Requirements
- Someone who has taken dexterous manipulation out of simulation and onto real robots. Grasping in simulation and grasping in real life are still far apart, and the only way to close that gap is hands-on experience. You've done that on multi-finger hands.
- 5+ years in robot manipulation on real hardware, with deployment to real robots not simulation
- Experience building end-to-end manipulation pipelines on dexterous / multi-finger hands: perception, grasp selection, execution, failure recovery
6D object pose estimation and grasp planning
- Inverse kinematics and arm / hand trajectory planning (e.g. MoveIt 2)
- Integration with hand drivers and motion control APIs, including force / tactile feedback
- ROS 2, C++, Python
Preferred
- Learning from demonstration and teleoperation data (imitation learning, VLA policies)
- Simulation for manipulation (Isaac Sim, MuJoCo)
- Humanoid or mobile manipulation experience
Hands are given.
What we're not looking for
- Experience that lives only in simulation or simulator training
- People who have only worked with simple grippers, not multi-finger hands
- People who want to design robotic hands from scratch
People who want to manage rather than build
People who want a remote or fixed-hours setup. We're in-person by choice and work happens next to the robots
