Role

ML engineers at Turing develop end-to-end machine learning models that power full autonomous driving, tackling a broad range of surrounding challenges—from model architecture and data quality to pipeline optimization, calibration, and training efficiency. You’ll build systems that learn from massive driving datasets, optimize for real-world performance, and deploy to physical vehicles.

Your work spans model architecture research, training pipeline optimization, data-centric approaches to improving model quality, evaluation frameworks, and deployment to on-vehicle systems. You collaborate with data engineers to understand driving data characteristics, with systems engineers to understand inference constraints, and with test drivers to gather real-world evaluation signals. This is applied ML research where papers become products.

What You’ll Do

Implement end-to-end autonomous driving models
Plan and manage data collection strategies and processes
Create and improve datasets
Implement and improve auto-labeling models
Perform camera and sensor calibration
Implement model training algorithms
Optimize model training code for performance and speed
Conduct model evaluation on real vehicles and manage experiments
Research, reproduce, and implement cutting-edge techniques from academic papers

What We’re Looking For

You have strong fundamentals in machine learning and can implement complex systems from research papers. You’re comfortable with large-scale training, distributed computing, and optimization. You think systematically about data quality and how it impacts model behavior.

Beyond strong technical skills, you ask good questions about what problems matter, why our current approaches work (and don’t), and what fundamental improvements are possible. You communicate clearly with cross-functional teams and contribute to a culture of rigorous technical thinking.

Tech Stack

Deep learning frameworks (PyTorch, TensorFlow)
Distributed training systems
Python and system-level optimization
Data processing and analysis tools
Model evaluation frameworks
Visualization and analysis tools

What Makes This Role Special

You’re working on the ML problems that directly enable full autonomous driving. Every model improvement translates to safer, more capable autonomous vehicles. You work with real driving data from real vehicles, seeing the direct connection between research and product. This is research-grade machine learning, but your models actually drive cars.

You’ll work closely with Kaggle Grandmasters and senior/staff-level engineers from leading technology companies. You’ll also have access to a large-scale GPU cluster approaching four-digit GPU counts—enabling you to run the kind of large-scale training experiments that most ML engineers never get to experience.

Key Qualifications

Strong foundation in machine learning fundamentals and deep learning
Experience implementing machine learning systems in production
Experience with large-scale training and distributed computing
Understanding of data quality’s impact on model performance
Experience with computer vision and/or sequence modeling for autonomous systems
Knowledge of or experience with camera and sensor calibration
Strong programming skills in Python
Ability to debug complex systems and analyze failure modes
Excellent communication and cross-functional collaboration

Cross-Functional Collaboration

With Software Engineers (MLOps)

You’ll collaborate on dataset creation, curation pipeline improvements, and acceleration of data processing and transformation. The central focus is improving pipelines to rapidly process growing volumes of data and accelerate the iteration cycle from data collection to model evaluation.

This is an opportunity to tackle the highly demanding technical problem of processing unstructured data at high speed. Building systems that rapidly spin the full loop—from data collection to experimental evaluation—dramatically multiplies the number of experiments you can run, directly boosting development efficiency.

With Software Engineers (Autonomous Driving Systems)

You’ll work together on understanding the characteristics of diverse real-world sensor data, processing and conversion, optimizing model performance on the AV system, and transmitting model output signals efficiently to other systems. Close collaboration is indispensable for maximizing ML model capabilities while maintaining system performance.

This is an opportunity to tackle the fundamental challenge of how to fully leverage real-world data for model training. Ensuring your model runs smoothly on real systems—with its outputs passed quickly and reliably to other modules—gives you a direct sense of contribution to realizing full autonomous driving.

With Infrastructure Engineers (GPU Cluster)

You’ll collaborate on verifying training job execution, identifying error causes, and accelerating model training. Together, you’ll build the foundation for efficiently utilizing the large-scale GPU cluster and accelerating ML development iteration cycles.

This is an opportunity to pursue acceleration of large-scale ML model training in a domain close to High-Performance Computing (HPC). You’ll directly benefit from infrastructure optimization—seeing faster training iteration as the tangible result of your collaboration.

With Reinforcement Learning Engineers

You’ll work on developing and improving simulators for autonomous driving, and evaluating models using simulators. The primary focus is on validating models before road deployment and generating edge-case data through simulation for use in training.

This is an environment where you can sharpen skills at the intersection of ML and reinforcement learning. Creating edge-case scenarios through simulators that are difficult to encounter on real roads—and contributing to more robust autonomous driving models—is an exciting challenge on the path to full autonomy.

With Autonomous Driving Test Drivers

Test drivers collect training data, control vehicles equipped with your deployed models, and provide real-time feedback on model behavior—directly contributing to the model improvement cycle. Test drivers are indispensable partners in autonomous driving AI development.