Machine Learning Engineer II - Autonomous Driving Training Infrastructure

May Mobility is transforming cities through autonomous technology to create a safer, greener, more accessible world. Based in Ann Arbor, Michigan, May develops and deploys autonomous vehicles (AVs) powered by our innovative Multi-Policy Decision Making (MPDM) technology that literally reimagines the way AVs think.

Our vehicles do more than just drive themselves - they provide value to communities, bridge public transit gaps and move people where they need to go safely, easily and with a lot more fun. We’re building the world’s best autonomy system to reimagine transit by minimizing congestion, expanding access and encouraging better land use in order to foster more green, vibrant and livable spaces. Since our founding in 2017, we’ve given more than 500,000 autonomous rides to real people around the globe. And we’re just getting started.

We’re hiring people who share our passion for building the future, today, solving real-world problems and seeing the impact of their work. Join us.

Job Summary

May Mobility is entering an exciting phase of growth as we expand our first-of-its-kind autonomous shuttle and mobility services across the nation. Launched in 2017 with a strong team of experienced roboticists and software engineers with decades of experience fielding robotic systems in the wild, May Mobility is looking to expand its team of robotics engineers with a background in robotics or autonomous vehicles.

We are seeking ML-Oriented Software Engineers with experience in robotics applications. As part of our Autonomous Driving ML team, you will use your knowledge of Software and ML concepts to design and operate pipelines that allow May’s Autonomous Driving stack to improve quickly and reliably at scale.

Essential Responsibilities

Own the process to transition Autonomous Driving ML models from concept to commercial scale.

Architect and operate data,  training and evaluation pipelines across cloud and cluster environments.

Design and maintain the data and metadata stores that back our training and evaluation workflows.

Skills and Abilities

Success in this role typically requires the following competencies:

Employ data and model parallelism training techniques for large data (>100TB) or large model (>100GB) applications

Architecting and operating containerized/pipelined ML Training workloads, including GPU scheduling/autoscaling, dataloader design and experiment tracking.

Building and maintaining CI/CD pipelines and infrastructure.

Working with relational and object stores, and high-throughput data formats for ML workloads.

Qualifications and Experience

• Candidates most successful in this role typically hold the following qualifications or comparable knowledge or experience:
• Required
• Bachelor’s or Master’s degree in Robotics, Computer Science or a related field with strong mathematical and engineering foundations.
• A minimum of 2 years building ML-oriented infrastructure, platforms, or distributed systems in production.
• Proficiency in C++, Python and PyTorch with experience in Linux environments.
• Familiarity with basic concepts in Machine Learning (training loops, basic operators and architectures)
• Desirable
• Familiarity with basic Perception and Planning concepts in Autonomous Driving.
• Proficiency in Go or Rust.
• Familiarity with ML orchestration and experiment tooling such as Ray, Kubeflow, Airflow, MLflow, or Weights & Biases.
• Familiarity with distributed training frameworks (PyTorch DDP/FSDP, DeepSpeed).
• Familiarity with data pipeline and storage technologies (Spark, Parquet, object storage, feature/metadata stores).
• Physical Requirements
• Standard office working conditions which includes but is not limited to: Prolonged sitting
• Prolonged standing
• Prolonged computer use
• Travel required? -  Low 5-10%