Mach9
ML Engineer, Product
$180K–$300K
San Francisco · Full-time · Posted Oct 1, 2026
- Experience
- Hands-on vision-model training or fine-tuning experience; the role is described as suitable for new graduates to approximately 3 years of experience.
- Education
- BS or MS in Computer Science, EE, Robotics, or a related field, or equivalent experience.
About the job
Builds and ships computer vision and machine-learning features that extract 3D objects and lines from LiDAR point clouds and imagery for Mach9’s AI-enabled CAD system.
What you'll do
- Develop extraction features end to end, from customer and product scoping through data strategy, model training, evaluation, and integration into Digital Surveyor.
- Adapt production models to new object classes, regions, and sensor types; build evaluation pipelines and iterate based on customer feedback and metrics.
- Use and build on existing model families and open-source tools for vision, point clouds, and geometry.
What they ask for
- BS or MS in Computer Science, EE, Robotics, or a related field, or equivalent experience.
- Strong coding abilities; preferably Python and Pytorch, and comfort working with a large codebase.
- Hands-on experience training or fine-tuning a vision model through work, internships, or projects.
- Working knowledge of 3D perception geometry, including coordinate systems, transforms, and projecting between images and point clouds.
Nice to have
- Prior use of open-source vision foundation models, such as the SAM family, DINO, or Grounding DINO.
- Experience with point clouds or LiDAR, or deploying models to production using ONNX, TensorRT, or batch inference on cloud GPUs.
Pay
$180K – $300K (base salary)
Tools
Python, Pytorch, Digital Surveyor, SAM family, DINO, Grounding DINO, Open3D, PDAL
AI focus
Develop and adapt computer-vision and machine-learning models to extract 3D objects and line features from dense LiDAR point clouds and imagery; train, fine-tune, evaluate, and integrate models into product workflows.
Model-written summary · the original posting is authoritative