Description
We build the embodied intelligence that moves real vehicles safely, and the ecosystem a billion machines will run on in the future. Every role here, whatever the team, plugs into that.
Vehicle Generalization Team
Vehicle Generalization makes it efficient, scalable, and predictable to deploy Wayve's autonomous driving stack across a wide range of vehicle platforms by making our core autonomy generalise as widely as possible. We work across pre- and post-training to improve how well our AI models generalise zero-shot to changes in perception,including cameras, intrinsics, field of view, lens distortion, and calibration,and embodiment, including vehicle dynamics, actuation limits, and rig geometry. We also own the Generalization Benchmark, Wayve's tool for measuring how models transfer across platforms. Our work goes into production programmes with major OEMs.
Responsibilities
- Design and train models that are robust to differences between vehicles, using approaches such as conditioning or augmentation for vehicle kinematics and camera parameters.
- Create benchmarks and metrics for cross-platform performance, translating results into clear decisions for model and platform teams.
- Help root-cause on-road issues such as late turns, lane-keeping, and steering problems. Work with controls, calibration, and OEM engineers to separate model problems from system problems and identify the highest-impact model improvements.
- Move research into production in collaboration with Driving Core and application engineering teams.
- Support customer programmes, with some travel.
- Mentor teammates and help raise the bar for ML engineering across the team.
Requirements
- You have several years of experience building and shipping ML systems in PyTorch and Python, including training at scale, debugging experiments, and rigorous evaluation.
- You have a strong robotics or automotive background in areas such as state estimation, sensor fusion, calibration, vehicle dynamics, or controls.
- You bring a systems perspective and can trace a problem through sensors, data, models, and actuation.
- You have experience working with real-world, multi-sensor driving or robot data.
- You communicate clearly across teams and are comfortable owning ambiguous problems from start to finish.
Nice to Have
- Experience with end-to-end or learned driving and planning policies, imitation learning, or reinforcement learning.
- Experience with platform bring-up or deployment across multiple vehicles.
- C++ skills and experience with on-vehicle or embedded deployment.
Benefits
- Salaries benchmarked against the market annually
- Meaningful equity
- Relocation support and visa sponsorship where applicable
- Hybrid working and access to vehicle workshops and labs
- Learning and development support
- Comprehensive location-dependent health, family, retirement, and wellbeing benefits