About the role
About the Company**
DiDi's autonomous driving unit was established in 2016 with the mission of developing Level 4 autonomous driving (AD) technology to make transportation safer and more efficient. In August 2019, the unit became an independent company, DiDi Autonomous Driving, dedicated to advanced AD R&D, product application, and business expansion. We believe integrating AD technology into a shared-mobility fleet will generate immense social value. By leveraging DiDi's specialized technology, operational expertise, and integrated ecosystem, we are positioned to build and operate a highly efficient, user-oriented autonomous fleet.
About the Role**
We are seeking a Software Engineer / Senior Software Engineer **to develop the next-generation map fusion and motion planning systems for our autonomous vehicles. In this role, you will bridge the gap between semantic HD maps, real-time sensor perception, and vehicle trajectory generation. You will design scalable software infrastructure, implement advanced geometric and deep learning frameworks, and develop the planning algorithms that enable our vehicles to navigate complex, dynamic environments safely and predictably.
Responsibilities**
- System Architecture**: Architect the data flow pipelines and APIs for map fusion, real-time map vectorization, and behavior/motion planning modules.
- Algorithm Deployment**: Design and deploy robust software frameworks that integrate offline High-Definition (HD) maps with online perception data to create a unified local environment model.
- Advanced Mapping Networks**: Implement and optimize state-of-the-art networks utilizing DETR-style, query-based vector decoding in bird's-eye-view (BEV) for online map element generation.
- Motion Planning & Optimization**: Design, implement, and validate core motion planning algorithms, establishing a tight feedback loop between vectorized map features, path generation, and trajectory optimization.
- Model Deployment Pipelines**: Own the end-to-end deployment pipeline for deep learning mapping models—from Python-based training and ONNX optimization to highly efficient runtime execution in C++.