Leads the team you would join
Owns the opening
On the team, can refer you
From your own Gmail and LinkedIn.
About the role
This is a full-time hybrid senior role in San Francisco for a machine learning engineer who builds and deploys 3D perception for autonomous construction equipment. The work covers sensor-fusion models that work in harsh job-site conditions and run on embedded edge hardware. Bedrock Robotics develops autonomy technology for heavy construction equipment. Candidates need five or more years taking deep learning models into production and a strong grasp of sensor calibration geometry.
What you’ll do
- Design and train models, such as BEV-based transformers, that fuse raw lidar and camera data for object detection and segmentation.
- Build perception that copes with occlusion from the machine's own arm or bucket, dust, snow, rain and heavy vibration.
- Optimize models for inference on embedded hardware and debug calibration drift and latency problems.
- Collaborate with other teams on representations for downstream uses.
What they’re looking for
- Five or more years taking deep learning models from research into production with PyTorch, TensorFlow or JAX.
- Deep understanding of SE(3) transforms, homogeneous coordinates and intrinsic and extrinsic calibration.
- Ability to project lidar points accurately onto image pixels.
- Practical experience with feature-level multimodal fusion architectures such as BEVFusion, TransFuser or PointPainting.
- Experience with modern transformer-based detectors such as DETR or PETR and temporal variants.
- Expert Python, with comfort in C++ or Rust systems code.
- Understanding of memory management and real-time constraints.
Nice to have
- Experience with occupancy grids, NeRFs or voxel representations for terrain mapping.
- Published work at conferences such as ICRA, IROS, CVPR, ECCV, ICCV, CoRL or RSS.
What’s on offer
- Hybrid working arrangement in San Francisco.
Skills
Questions about this role
Is this role remote?
No. It is hybrid and based in San Francisco.
How many years of experience do I need?
At least five years taking deep learning models from research into real-world production.
Which frameworks are needed?
PyTorch, TensorFlow or JAX, with Python and some C++ or Rust.