Autonomous Vehicle Intern, Data Platform / Data Mining

Key Skills

AI Robotic systems, C++, CUDA kernels, Data Mining, Data Visualization, Machine learning techniques, MySQL, NeurIPS, PostgreSQL, Python Programming, PyTorch, SQL, Tableau, TensorFlow

Job Description

At Motional, we believe in Data Science and Artificial Intelligence: they are the pillars of our world-class, machine learning-based, autonomous driving stack. By joining the Data Mining team, you will directly contribute to making roads safer, by helping us harness the potential of our collection of hundreds of hours of driving scenarios.

In particular, you will participate in the improvement of our perception stack, by identifying valuable training & evaluation scenarios for our lidar-based perception models. You will take part in the definition of strategies to select and use training data via the analysis of pseudo ground-truth labels (auto-labels) and attributes that get extracted by our data & mining pipelines. You will experiment with your own approaches and criteria, and validate them via training experiments. The internship will take place in our Singapore office and we expect a full-time internship period of at least 6 months. We offer flexible working hours and allow for remote work within Singapore.

What you’ll be doing:

• Work with a data-oriented team that develops and improves various machine-learning algorithms for perception, prediction and planning.
• Analyze the fidelity and limitations of our auto-labeling system from a statistical standpoint.
• Generate data visualization dashboards to measure the quality of the data samples that are at our disposal.
• Define learning strategies and data queries to select the most reliable and informative samples to use for perception model training.
• Carry on training experiments with the selected data and perform quantitative and qualitative error analysis against real-world perception failure modes.
• Contribute to patents and/or scientific publications.

What we’re looking for:

• Currently pursuing or recently completed a Master's or Ph.D. degree in Computer Science, Applied Mathematics, Statistics, Robotics or related fields.
• Prior experience in training and analyzing neural networks for at least one of the following applications: Object detection, Instance/Semantic segmentation, Sensor fusion.
• Proficiency in Python and at least one Machine-Learning framework (PyTorch, Tensorflow) in a Linux environment.

Bonus points (not required):

• Proficiency in SQL and relational databases (PostgreSQL, MySQL, etc.).
• Experience with data visualization tools (Tableau, Power BI, Redash, etc.).
• Knowledge about point-cloud processing via geometrical and AI-based approaches.
• Industrial experience in software development.
• Familiarity with C++, CUDA, PyTorch Lightning, Hydra etc.
• Experience in training machine-learning algorithms on the cloud (e.g. AWS).
• Proven track record of publications in relevant conferences (CVPR, ICML, NeurIPS, ICRA, IROS, IV etc.).

Why you should join us:

Be part of a company that works on making a difference, by pushing forward the limits of the current mobility solutions! Join the team that released the nuScenes and nuPlan frameworks at Symbiosis Tower, in the center of the futuristic Fusionopolis area, a regional research and development hub. A new office space with standing desks, unlimited snacks and beverages, ample entertainment including video games, a variety of retail and dining offerings nearby, and an ideal location beside mass transit is waiting for you!