Lead Machine Learning Engineer

About the role

Serve Robotics is building autonomous sidewalk delivery robots that operate in major US cities. As a Lead Machine Learning Engineer you will design and scale large‑scale training systems for multimodal robotics data, enabling high‑performance autonomy models. You will work closely with researchers and infrastructure teams to accelerate model iteration and maximize GPU utilization.

Key responsibilities

• Design and maintain training pipelines that process petabyte‑scale video, point‑cloud and sensor datasets across large GPU clusters.
• Identify and eliminate bottlenecks in data loading, preprocessing, model computation and inter‑node communication to improve training efficiency.
• Develop and refine neural network architectures and loss functions for high‑dimensional sequential sensor data.
• Configure, monitor and maintain distributed training jobs, ensuring fault tolerance and optimal resource usage.
• Build tools and workflows for experiment tracking, data augmentation and rapid model iteration.

Required profile

• Master’s or PhD in Computer Science, Robotics, Machine Learning or a related field.
• Minimum 5 years of professional experience developing, training and deploying ML models in production.
• Hands‑on experience with multi‑GPU or multi‑node distributed training frameworks.
• Strong programming skills in Python for building data pipelines and training workflows.

Required skills

• Python
• Distributed training frameworks (e.g., PyTorch Distributed, TensorFlow)
• GPU cluster management
• Neural network architecture design
• Multimodal data processing (video, LiDAR point clouds, telemetry)