AI/ML Engineer - Model Inference

General Motors•Sunnyvale, CA

2d•$117,700 - $221,400•Remote

About The Position

The Team Cola is part of GM’s autonomous vehicle effort, focused on helping teams discover, understand, and curate high-value data from large-scale real-world sensor streams. The team sits at the intersection of machine learning, data infrastructure, and developer productivity, building systems that make it easier to search for important scenarios, prepare training-ready data, and support fast iteration across perception and evaluation workflows. Our goal is to make world understanding scalable, practical, and cost efficient for embodied AI systems. We believe the next generation of autonomy and robotics depends not only on stronger models, but also on better infrastructure for turning massive volumes of multimodal data into reusable signals, searchable artifacts, and high-quality evaluation loops. That means building systems that can operate at industrial scale while preserving the flexibility to adapt quickly to new questions, new edge cases, and new model capabilities. A core idea behind how we work is EMWU, or Efficient Multi-Tier World Understanding. At a high level, EMWU is a cost-aware approach that first performs the cheapest reusable work, such as detection, featurization, and retrieval, and then applies deeper reasoning only where it adds meaningful value. This operating model reflects how Cola engineers think: build durable intermediate artifacts, design for scale from the start, and balance quality, speed, and cost instead of optimizing any one of them in isolation.

Requirements

BS, MS, or PhD in Computer Science, Electrical Engineering, Robotics, or a related technical field, or equivalent practical experience
Experience building production data processing or machine learning pipelines at scale
Experience with featurization, embedding, inference, or retrieval systems for vision or multimodal workloads
Strong understanding of computer vision models and the practical challenges of deploying them in production environments
Experience evaluating machine learning systems using clear metrics, experiments, and regression safeguards
Proven ability to work hands-on in fast-moving environments with incomplete information
Strong ownership mindset, sound technical judgment, and the ability to drive execution through ambiguity

Nice To Haves

Experience with world models or large-scale world understanding systems
Experience with simulation workflows or synthetic data systems
Experience with vector search, approximate nearest neighbor retrieval, or large-scale embedding infrastructure
Experience working on embodied AI, autonomous systems, or safety-critical machine learning applications

Responsibilities

Design, build, and productionize data processing and featurization pipelines for large-scale multimodal data
Improve inference frameworks for computer vision and multimodal models, with a focus on reliability, extensibility, and operational simplicity
Drive scalability and cost efficiency across the end-to-end pipeline, including compute utilization, throughput, storage, and query performance
Work closely with partners across machine learning, infrastructure, and evaluation to deliver systems that support both rapid experimentation and production use
Develop and refine evaluation methods for model quality, retrieval quality, and system-level performance
Help shape technical direction through strong execution, thoughtful tradeoff analysis, and clear engineering judgment
Take ownership of ambiguous problem spaces, define practical paths forward, and move quickly from prototype to production
Operate with urgency and a strong bias toward execution velocity while maintaining a high bar for engineering quality