Software Engineer - Machine Learning, Planner/Behavior Prediction

Didi autonomous driving is innovating the way we build our self-driving cars.  We leverage our ride-sharing network to collect large amounts of data, and improve our algorithms with the data. If you are interested in solving exciting and challenging problems, and are enthusiastic about autonomous driving, please join us!  

Responsibilities:

• Design data-driven algorithms to improve the safety, generalization and reliability of self-driving algorithms, including: framing the open-ended real-world driving problems into well-defined ML problems; building training / evaluation pipelines to solve the problem, and landing them in real world driving.
• Develop and apply cutting-edge ML approaches (deep learning, reinforcement learning, imitation learning, decision trees, etc) to these problems.
• Gracefully handling many complex situations involving social aspects (merging, negotiating narrow roads, etc), while dealing with noisy, uncertain, and incomplete information.
• Collaborate with other teams, including the ML infrastructure, perception and systems engineering teams.

Qualifications:

• MS/PHD in Computer Science, Robotics, similar technical field of study, or equivalent practical experience
• Experience solving problems using Machine Learning with Tensorflow/PyTorch or equivalent tools
• Ability to collaborate within and across teams
• Strong experience programming in C++ or Python with robust and efficient code

Preferred:

• Experience specifically with applying Machine Learning to solve motion planning or related robotics problems
• Experience with Deep Learning Models (e.g. RNN/LSTM, CNN, VAE, GAN, etc.)

 

About the Company:   Didi Chuxing (“DiDi”) is the world’s leading mobile transportation platform. We’re committed to working with communities and partners to solve the world’s transportation, environmental, and employment challenges by using big data-driven deep-learning algorithms that optimize resource allocation.   Didi Chuxing’s Autonomous-Driving team was established in 2016, and has grown to a comprehensive research and development organization covering HD mapping, perception, behavior prediction, planning and control, infrastructure and simulation, labeling, hardware, mechanical, problem diagnosis, vehicle modifications, connected car, and security, among others. We’re developing and testing self-driving vehicles in China and the United States.   In August 2019, DiDi upgraded its autonomous driving unit to an independent company to focus on R&D, product application, and business development related to self-driving technologies. The new company will integrate the resources and technology of DiDi’s platform, continue to increase investment in R&D, and deepen collaboration with auto industry partners.