Senior, ML Engineer - VLM

Ann Arbor, MI · Remote · full time · Senior

$177,300 – $212,800

Listed on Torc Robotics’s own careers site. You apply with them directly — we never stand between you and the employer.

What this role is

This role leads the development of high-quality training datasets for Torc's vision-language and end-to-end autonomous driving models, converting petabytes of fleet sensor data into labeled datasets that drive model performance. It suits experienced ML engineers who combine deep expertise in computer vision, multimodal models, and distributed data systems with the judgment to own a critical technical pipeline and mentor others.

Our summary, not Torc Robotics’s wording. The full posting is on their site.

Skills this role names

What they ask for

Required

Computer vision and deep learning with model training experience plus at least two of: 2D/3D object detection, tracking, sensor fusion, semantic segmentation, BEV, or depth estimation
Hands-on experience with vision-language models, open-vocabulary recognition, dense captioning, or semantic embeddings applied to perception data
Experience building targeted datasets that measurably improve downstream model performance and processing large-scale Parquet data
Distributed ML and data frameworks such as PyTorch, Lightning, Ray, Spark, or equivalent
Scaled MLOps and tooling including experiment tracking, model registry, MLflow or Weights & Biases
Strong Python software development, cloud-based development environments, CI systems like GitHub Actions, and Docker
Bachelor's degree in Computer Science, Robotics, Electrical Engineering, or related field plus 6+ years of relevant experience, or Master's degree in related field plus 3+ years of experience

Nice to have

Familiarity with VLM/VLA or end-to-end driving models, trajectory and action grounding, or reasoning-trace datasets
Experience with auto-labeling foundation models, segmentation, open-vocabulary detectors, or VLM/LLM-driven data engines
High-throughput model serving tools like vLLM or SGLang for batch auto-labeling and inference at scale
Semantic inference and retrieval including attribute mapping, semantic search, and vector databases like LanceDB
Knowledge of AV data standards and tooling including Pegasus layers, robotics formats like ROS bags and MCAP, and columnar storage optimization
Cloud development and orchestration with Terraform and AWS managed services (S3, ECS, Lambda, DynamoDB, Step Functions, Athena)
Data visualization tools such as Foxglove, FiftyOne, three.js, or OpenGL
Publications in top-tier CV/AI/Robotics venues or experience with closed-loop evaluation frameworks