Senior Software Engineer, Perception (Robotics)

About the role

Grab's Robotics Technology team is building urban embodied AI, from in-house hardware and control systems to real-world fleet testing. As a Senior Perception & Prediction Engineer, you will turn raw multi-sensor data into a reliable, real-time and predictive representation of the world, working across a modular perception stack and learning-based approaches. You will deepen multi-object tracking, generic open-set world understanding and motion prediction, and help productionise end-to-end temporal perception-prediction models and VLA / embodied foundation models. The role reports to the Senior Principal Perception & Prediction Engineer and is based onsite at a Grab office.

Key responsibilities

• Develop and improve multi-object tracking, including data association, Bayesian filtering (Kalman / EKF / UKF), track lifecycle and ID stability, and evolve it toward learning-based tracking.
• Build generic, open-set world understanding such as class-agnostic obstacle detection, occupancy grids / networks / BEV occupancy and learned generic-object branches.
• Build motion prediction for pedestrians, cyclists and vehicles, including multi-modal trajectory forecasting, interaction-aware prediction, occupancy flow and uncertainty estimation.
• Develop and evaluate end-to-end temporal perception-prediction approaches while keeping modular baselines and production fallbacks.
• Explore VLM / VLA and embodied foundation models, then distil, quantise or adapt useful capabilities for on-robot deployment.
• Strengthen multi-sensor and temporal integration across camera, LiDAR, radar and IMU, and define quality gates for tracking, prediction and safety.
• Partner with Detection, Prediction, Planning, Data & Infra and Integration teams to move capabilities from research to simulation, NVIDIA Orin / TensorRT deployment and production testing.

Required profile

• At least 3 years of hands-on experience shipping or operating perception and/or prediction systems for robotics, AV or ADAS, with strong autonomous-driving experience.
• Rich experience across classical/geometric and learning-based approaches, with sound judgement on modular, end-to-end and hybrid designs.
• Proven depth in multi-object tracking and hands-on experience with generic, open-set perception.
• Hands-on depth in at least one modern temporal area (motion forecasting, occupancy flow, streaming BEV, learned world models).
• Experience navigating noisy production environments through reproducible experiments, metrics and log-driven root-cause analysis.

Required skills

• Proficiency in C++ and Python, with production-quality code and real-time embedded awareness.
• Fundamentals of geometry and multi-sensor systems, including calibration and alignment diagnostics.
• Good to have: BEV detection and sensor fusion (e.g. BEVFusion / LSS), ROS 2, NVIDIA Orin, TensorRT, DeepStream, and foundation-model adaptation (LoRA, distillation, quantisation).

What we offer

• Term life insurance and comprehensive medical insurance.
• Parental and birthday leave, plus volunteering leave.
• A confidential assistance programme and flexible work arrangements.