3D Machine Learning Engineer

Job description

Job responsibilities

• Design and implement scalable machine learning pipelines for large-scale 3D spatial data processing for point cloud analysis, object detection, segmentation, and scene understanding.
• Train, optimize, and deploy deep learning models using PyTorch, TensorFlow, or equivalent frameworks on cloud platforms such as AWS (e.g., SageMaker, EC2).
• Collaborate with software and systems engineers to integrate models into production environments and continuously improve inference pipelines.
• Analyze diverse sensor inputs, including RGBD imagery, LiDAR point clouds, 360 photos, audio, and Building Information Models (BIM).
• Work closely with the labeling and data operations teams to define robust data annotation strategies and ensure high model performance and generalization.

Job requirements

• Bachelor’s or Master’s degree in Computer Science, Machine Learning, Robotics, or a related technical field.
• 2+ years of hands-on industry experience developing and deploying machine learning systems for 3D point clouds, perception, or spatial understanding tasks.
• Strong background in 3D machine learning, with experience in deep learning for point clouds, multi-view fusion, or geometric learning.
• Strong expertise in Python and deep learning frameworks: PyTorch, TensorFlow, or similar.
• Familiarity with OpenCV and PCL (Point Cloud Library) for classical computer vision and 3D data preprocessing.
• Experience training, evaluating, and deploying ML models using cloud infrastructure (e.g., AWS, SageMaker) and containerized workflows.
• Solid understanding of the end-to-end ML lifecycle, including experiment tracking, reproducibility, model versioning, and optimization for production.
• Proven ability to work in fast-paced, interdisciplinary teams across software, ML, and product teams.
• Experience working with BIM data, digital twins, or construction-related sensor data.
• Background in geometric deep learning, 3D mesh analysis, GIS systems, or structured scene representations.
• Familiar with MLOps pipelines using Ray, SageMaker, MLflow, or Kubeflow.
• Strong foundation in geometric computer vision, robotics, or algorithmic 3D reasoning.
• Exposure to graph neural networks, geodesic computations, or neural implicit representations (e.g., NeRF, Occupancy Networks).
• Deep experience with point cloud and graph learning frameworks such as Open3D-ML, Torch-Points3D, PyG, or MMDetection3D.
• Experience building custom modules for SparseConvNet or 3D transformers.