Senior ML Platform Engineer

About the role

42dot is hiring a Senior ML Platform Engineer for its AD ML Platform team, which builds the core data platform and ML training and evaluation platform for autonomous driving algorithms. The team develops a distributed, scalable data platform for millions of scenes and high-performance data-serving SDKs for ML model training and evaluation, improving the efficiency of the ML model development lifecycle.

Key responsibilities

• Set the technical strategy and oversee development of a high-scale, reliable data platform that manages, visualizes and serves large datasets for ML model training and validation.
• Build the data lakehouse for autonomous driving scene datasets, including sensor data, calibration data and annotation data.
• Drive development of the Autonomous Driving Data SDK, including scene data search and dataset preparation and loading.
• Analyze performance bottlenecks across data processing pipelines, from data processing and search latency to test procedure coverage.
• Bootstrap and maintain infrastructure for data processing pipelines, databases, data lakehouses and data serving.
• Collaborate with ML algorithm, ML application and cloud infrastructure teams to align ML platforms with the overall autonomous driving system architecture.

Required profile

• Bachelor's degree or higher in Computer Science, Engineering, Robotics or a similar technical field.
• Minimum of 7 years of experience in Data Engineering or ML Platform roles.
• Leadership and communication skills, with a demonstrated ability to lead technical projects.

Required skills

• Expert-level Python and solid experience building Python SDKs.
• Solid experience with databases such as MongoDB and PostgreSQL.
• Understanding of modern AI frameworks such as PyTorch and TensorFlow, especially distributed data loaders for model training.
• Data pipeline orchestration with Databricks Workflows or Apache Airflow, integrated with machine learning models.
• Data warehouse (e.g., Hive) or lakehouse (e.g., Delta Lake) technologies, and Apache Spark or other big data engines.
• Preferred: autonomous vehicle sensor data (LiDAR, camera, radar), ML model training lifecycle, data governance and security, and large models such as VLM.

The interview process includes application screening, a coding test, two interviews (about 1 hour and about 3 hours) and an offer discussion; a 3-month probation may apply. Applications are made through the company's recruitment channel.