ML Engineer
Encube
Encube is a platform where hardware teams review 3D designs and engineering drawings to build better products faster using AI that understands drawings and 3D models. The role focuses on training models that connect 2D drawings to 3D models, building agents that generate CAD and drawings, measuring model quality, and shipping solutions to customers; the post also lists an ML Ops Engineer role focused on GPU infrastructure and reproducibility.
What you’ll do
- Train models that connect 2D drawings to 3D models
- Build agents that generate CAD and drawings
- Figure out how to measure if models are actually good
- Ship models to real customers
- Make experiments reproducible
- Keep costs down and customer data safe
- Remove bottlenecks slowing the ML team down
What they’re looking for
- Experience with PyTorch
- Bonus if you've worked with 3D, CAD or graph neural nets
- Experience with SkyPilot, MLflow, PyTorch for ML Ops
- Experience with on-prem and AWS GPUs
Skills
- PyTorch: PyTorch, train models
- 3D: 3D, 3D models
- CAD: CAD, generate CAD, drawings
- Graph Neural Networks: graph neural nets
- SkyPilot: SkyPilot
- MLflow: MLflow
- AWS GPUs: AWS GPUs, on-prem GPUs
The post
Encube | ML Engineer, ML Ops Engineer | Stockholm, Sweden | ONSITE | Full-time
Encube is a platform where hardware teams look at their 3D designs and engineering drawings together and make better products faster. We're building AI that understands those drawings and 3D models, and can eventually create them. Evaluations are a massive priority, and we need to consistently demonstrate how our models are better than frontier VLMs.
ML Engineer: train models that connect 2D drawings to 3D models, build agents that generate CAD and drawings, figure out how to measure if they're actually good, and ship them to real customers. PyTorch. Bonus if you've worked with 3D, CAD or graph neural nets.
ML Ops Engineer: on-prem and AWS GPUs, make every experiment reproducible, keep costs down and customer data safe, and get rid of whatever slows the ML team down. SkyPilot, MLflow, PyTorch.