Jobs at XPeng Motors

Develop, evaluate, and deploy reinforcement learning methods for autonomous vehicle planning. Design real-time model architectures, build offline data-driven ML infrastructure, analyze fleet and field-testing data, optimize and deploy planning models, and collaborate with perception, software, and AI teams to deliver production autonomous driving systems.

Research, implement, and evaluate deep-learning methods for legged locomotion and whole-body control in humanoid robots. Develop end-to-end robot motion controllers using reinforcement and imitation learning, design and run experiments to address sim-to-real challenges, and integrate state-of-the-art research.

Intern will design and optimize software to deploy large-scale AI models on production vehicles, profile and improve inference across compute, memory, and I/O, reduce latency and power use on embedded automotive platforms, and deliver production-ready, fleet-scalable performance optimizations.

Design and operate petabyte-scale end-to-end data pipelines for autonomous driving: onboard upload, cloud preprocessing, dataset production, and model training input. Build data cleaning, annotation QA, lineage, versioning, metadata management, and high-throughput distributed processing. Optimize I/O, memory, transmission, and containerized deployments; support cross-team collaboration and algorithm iteration.

Lead development and productionization of LLM quantization and deployment pipelines (PTQ, QAT, mixed-precision). Implement robust Python production code, build export/calibration/benchmarking workflows, collaborate across research, field-testing, vehicle software, and training infrastructure teams, and curate evaluation datasets and metrics to validate model accuracy, latency, and numerical consistency for on-vehicle inference.

Lead development and productionization of LLM quantization and deployment pipelines (PTQ, QAT, mixed-precision, INT8/FP4), build robust export/calibration/benchmarking/validation tooling, collaborate with research, field-testing, simulation and vehicle software teams, curate evaluation datasets and metrics, analyze numerical errors and performance trade-offs, and develop orchestration workflows for on-vehicle inference and model sign-off.

Develop and optimize end-to-end data pipelines for autonomous driving, ensure data quality, and support model training through efficient data management processes.

Design, build, and operate a production AI engineering platform that connects agents to engineering systems. Build reliable distributed services, automation, agent lifecycle and safe action execution, semantic data layers, and observability. Lead customer onboarding, gather requirements, debug production issues, and drive adoption while establishing evaluation and safety guardrails.

Lead cross-functional programs to build and operate an autonomous-driving training data lake: define roadmap, coordinate data/ML/infrastructure/annotation teams, set KPIs for data quality and pipeline reliability, manage dataset generation/versioning/lineage, identify and resolve technical bottlenecks, and improve engineering and operational processes to deliver high-quality training data at scale.

The Senior Staff Physical AI Data Algorithm Engineer is responsible for designing and optimizing a vehicle-cloud data closed-loop architecture, managing data toolchains, and ensuring high-quality data standards throughout the model development cycle.

Lead the architecture design and R&D in synthetic data and simulation systems for autonomous driving and robotics, ensuring high-quality data generation and closed-loop simulations.