Back to Careers
Spacecraft Perception Researcher
Full-time
·
Dubai or remote
·
Research
The Mission
Own the number the company stands on: raising single-attempt capture probability from 42% to 77% by taking monocular pose estimation of tumbling spacecraft to sub-degree accuracy on SPEED+ and beyond, then closing the sim-to-real gap on our lab testbed.
What You’ll Own
Lead 6-DoF pose estimation research for non-cooperative, tumbling targets under adversarial lighting
Train and evaluate on SPEED+ and synthetic pipelines; publish where it strengthens the company
Build the domain-randomization and data pipeline with our AI engineer
Integrate perception outputs with the world-model tumble predictor and GNC stack
Define and hit the Phase 0 milestone: sub-degree pose error, demonstrated in closed loop
What We Need
PhD or equivalent research record in vision-based pose estimation, 6-DoF estimation, or spacecraft-relative navigation
Familiarity with SPEED/SPEED+ benchmarks or comparable spacecraft imagery work
Strong PyTorch and modern vision architectures (ViT, keypoint, direct regression methods)
Evidence of taking research from paper to working system
Nice to Have
Sim-to-real transfer or domain adaptation publications
Experience with VLMs for structured visual tasks
Robotics lab experience with hardware-in-the-loop testing