Role

HILOS is building an AI-driven design platform to enable intuitive creation and modification of manufacturable 3D geometry. We are seeking a machine learning engineer with deep expertise in 3D geometry processing and learning-based methods to join our team. Your work will directly impact the way we generate, edit, and reverse-engineer CAD-compatible shapes for digital manufacturing workflows.

The ideal candidate will bring experience in learning with 3D geometric data (point clouds, meshes, SDFs) and a strong understanding of how these models integrate with or inform CAD-based design systems. Familiarity with approaches that bridge neural models and CAD representations—such as LLMs for CAD code or geometry-to-code systems (e.g., CAD-Recode)—is a plus, though not required.

Responsibilities

• Research, implement, and optimize deep learning methods for 3D geometry understanding, generation, and reconstruction using both explicit and implicit representations.
• Develop pipelines that translate sparse or high-level input (e.g., sketches, point clouds, prompts) into structured and editable CAD representations.
• Explore reverse engineering pipelines that extract parametric or code-based design primitives from point clouds or mesh scans.
• Build scalable systems for training, evaluating, and deploying 3D ML models, with attention to manufacturability, surface fidelity, and performance.
• Read and replicate state-of-the-art research papers, evaluating their applicability to internal problems and reimplementing them as needed.
• Work closely with computational design and software teams to integrate ML outputs into CAD-driven workflows.
• Contribute to the architecture of an AI-first 3D design platform that prioritizes ease of use and real-time interactivity.

Preferred Experience

• Strong background in 3D deep learning, including experience with:
  • PointNet++, KPConv, sparse convolutional networks, 3D diffusion, or transformer-based models.
  • Implicit representation models (e.g., DeepSDF, OccNet, VolRecon).
  • Mesh or parametric surface reconstruction from learned representations.
• Familiarity with geometry-aware ML libraries and tools (e.g., PyTorch3D, Kaolin, Trimesh, Open3D).
• Ability to interpret and implement cutting-edge research from academic papers, with strong debugging and evaluation skills.
• Understanding of CAD principles and parametric modeling. Experience with integrating ML models into CAD systems (e.g., Rhino, Fusion 360, Onshape) is a strong plus.
• Interest or experience in CAD-code generation from ML models (e.g., LLMs that produce OpenSCAD or Grasshopper scripts).
• Proficiency in Python and deep learning frameworks such as PyTorch.
• Experience building end-to-end ML systems, including training pipelines, evaluation metrics, and cloud-based deployment.

Qualifications