Machine Learning Engineer II

About the role

Driving Measurement That Shapes Cities

May Mobility operates at the intersection of technology and urban life, reimagining how people move. Headquartered in Ann Arbor, Michigan, we design and deploy autonomous vehicles powered by Multi-Policy Decision Making (MPDM) technology. Our systems do more than navigate; they bridge public transit gaps, move people safely and enjoyably, and reduce congestion. By encouraging thoughtful land use, we help create greener, more vibrant, and accessible communities for everyone. Since 2017, we have provided over 500,000 rides to real passengers worldwide.

We are expanding our first-of-its-kind autonomous shuttle services across the nation and seeking ML-Oriented Software Engineers to join our mission. You will join the Autonomous Driving ML team, where your expertise in robotics and autonomous vehicles will directly influence how our technology performs in the real world.

Designing Measurement Frameworks

Your primary role is to design the measurement frameworks that define autonomous performance. You will own metric definitions and construct pipelines for offline model evaluation, simulation analysis, and on-road performance tracking. The goal is to translate complex operational data into clear, actionable insights. You will work closely with product managers and city partners to ensure our analysis reflects real-world community needs. Your measurements will guide vehicle behavior and influence public trust in autonomous mobility.

Building Robust Evaluation Systems

You will be responsible for building and maintaining test, regression, and hillclimbing suites. These systems act as critical gates for model and stack releases, ensuring reliability before deployment. When regressions occur, you will trace failures to their root causes within the perception and planning stacks. This work demands rigorous attention to detail and a commitment to safety. Your analysis will directly prevent issues from reaching our riders.

Driving Model Improvement Through Data

A core part of your contribution involves using data to improve the models themselves. You will perform loss analysis and error mining to identify weaknesses in system behavior. Based on these findings, you will design data balancing and curation strategies for both training and evaluation sets. You will reshape evaluation criteria to address long-tail scenarios that impact safety and performance. This iterative process is essential for building a robust autonomy stack.

Analysis, Simulation, and City Alignment

You will analyze simulation and on-road data slices to surface rare but critical long-tail scenarios. The insights you generate will feed into our shipping analysis dashboards. These dashboards align model metrics with city operations and rider experience signals. You will also validate data curations through comparisons that respect the constraints of operating in live urban environments. Coordination with the MPDM teams ensures metric coherence across planning, prediction, and control modules.

Collaboration and Impact

Your work shapes how communities experience safe and accessible streets. By aligning autonomy features with public goals around access and emissions, you help transit agencies achieve their objectives. You translate messy operational data into narratives that city leaders and partners can understand. This collaboration ensures that technical metrics support tangible civic outcomes. Your role connects complex software systems to the people who rely on them.

Qualifications and Experience

To succeed in this role, you need a strong educational background. A Bachelor's or Master's degree in Robotics, Computer Science, Statistics, or a related field is required. You must bring a minimum of two years of experience building evaluation, metrics, or data analysis systems for machine learning in production. Proficiency in Python is essential, along with NumPy, Pandas, or similar dataframe tools. You must work comfortably in Linux environments.

A solid understanding of machine learning fundamentals is necessary. This includes knowledge of losses, train/eval splits, and common failure modes. You should also understand core autonomy concepts, particularly perception and planning within vehicle control stacks.

Preferred Skills

Fluency in Go or C++ is a significant advantage for this position. Experience with experiment tracking and evaluation tooling is also valuable. Familiarity with systems like MLflow or Weights & Biases, or in-house equivalents, is desirable. These tools help you manage the complexity of multiple experiments and model versions.

The Tools You Will Use

Your daily work will involve the May Mobility Multi-Policy Decision Making stack. You will utilize autonomous vehicle sensors and control systems. Proficiency in Python, NumPy, and Pandas will be central to your daily tasks. Experience with Go or C++ is helpful for deeper system integration. You will also use MLflow and Weights & Biases to track experiments and analyze results.

Practical Information

Please confirm all details regarding location and compensation on the official application page. This role is open to candidates located anywhere in the USA. It is a full-time position designed for individuals who thrive on solving real-world problems and seeing the direct impact of their work. Join us in building the future of urban mobility today.

What you'll do

• Meet the bar Practical notes