Senior Planning & Prediction Engineer — Autonomous Driving (L4 Heavy-Duty Trucks)
Team: Autonomous Driving · Planning & Prediction
Location: Singapore
Employment type: Full-time
About the Role
We are building full-stack Level 4 autonomous driving systems for heavy-duty trucks in highway and trunk logistics, and we are standing up our overseas planning team. This is an early, high-ownership role: you will help architect the prediction, decision, and motion-planning stack for our global product.
Heavy trucks are not passenger cars. A loaded tractor-trailer has long stopping distances, limited acceleration, articulated dynamics, and real rollover and jackknife limits — so planning has to think much further ahead and respect physics that a robotaxi can ignore. If you want to make high-stakes decisions in dense highway traffic, and see them run on real freight lanes at scale, this is the place.
What You’ll Do
- Architect and develop the prediction, behavior, and motion-planning stack for L4 highway trucking.
- Build multi-agent trajectory prediction and intent/interaction modeling for surrounding traffic.
- Drive the stack toward learned and end-to-end planning — including joint prediction-planning and data-driven decision-making — and help lead the shift from modular pipelines toward end-to-end autonomous driving.
- Design behavior and decision-making for highway maneuvers: lane keeping and changes, merges, on/off ramps, gap selection, and interaction with cut-ins and merging vehicles.
- Develop motion planning and trajectory optimization that respect heavy-truck kinematics and dynamics (mass, articulation, stability limits) while balancing safety, smoothness, and efficiency.
- Account for fuel/energy efficiency in planning — predictive, eco-driving behavior over highway terrain.
- Train, optimize, and deploy models on automotive-grade compute (e.g., NVIDIA DRIVE Orin/Thor) under hard real-time constraints.
- Tackle highway long-tail decision scenarios and build planning with safety in mind, working with our safety team where relevant.
- Collaborate closely with perception, control, mapping, data, and platform teams.
What We’re Looking For
- MS or PhD in Computer Science, Robotics, EE, Controls, Applied Math, or equivalent practical experience.
- Strong fundamentals in motion planning — search-, sampling-, or optimization-based methods (e.g., A*/lattice planners, RRT, iLQR, MPC-based planning).
- Experience with behavior/decision-making (state machines, behavior trees, POMDPs, or learning-based decision policies).
- Experience with trajectory prediction and multi-agent/interaction-aware modeling.
- Hands-on end-to-end/learned planning background. Direct experience with learned or end-to-end planning and prediction — e.g., imitation- or RL-based planning, learned cost models, transformer-based motion planning, or joint prediction-planning.
- Solid grasp of vehicle kinematics and dynamics and how they constrain feasible trajectories.
- Proficient in C++ and Python, comfortable working under real-time constraints.
Nice to Have
- Production autonomous-driving/ADAS experience, ideally shipped to vehicles.
- Commercial-vehicle or heavy-truck AV experience.
- Deep-learning prediction or planning (transformers, graph networks, learned cost models).
- Optimization and numerical methods (QP/NLP solvers, real-time optimization).
- Eco-driving/predictive energy-efficient planning.
- NVIDIA DRIVE platform (Orin/Thor) and embedded experience.
- Familiarity with functional safety (ISO 26262) or SOTIF (ISO 21448).
- Top-venue publications (CoRL/RSS/ICRA/NeurIPS/CVPR).
Why Join Us
- L4, not L2 — full self-driving for commercial trucking, not driver-assist.
- Real deployment at scale on commercial freight lanes, with a fast feedback loop from road to model.
- A global, founding-stage team where your architectural decisions set the direction.