AI/Computer Vision 2D Blueprint to 3D Geometry Extraction & Real-Time Object Detection

Job Description: We are looking for a highly skilled Computer Vision Engineer to help build the core engine of a construction management SaaS platform. The project has two primary technical pillars: 2D-to-3D Reconstruction: Developing a pipeline to take 2D architectural blueprints (PDF/images), extract structural geometry (walls, openings, dimensions), and convert them into a structured 3D model. Real-Time Site Monitoring: Implementing real-time object detection using YOLOv8/v10 or Detectron2 to identify construction elements (pipes, windows, doors, ceilings) from site-uploaded photos to track project progress. Key Responsibilities: Develop algorithms to parse and binarize 2D blueprints, identifying room boundaries and structural symbols. Implement geometry extraction logic to transform 2D coordinates into 3D mesh or CAD-compatible data. Train and fine-tune object detection models (YOLO/Detectron2) on construction-specific datasets. Integrate model inference into a scalable API (FastAPI/Flask) for our web-based platform. Optimize models for real-time performance and accuracy in messy, real-world construction site conditions. Technical Requirements: Deep Learning Frameworks: Expert proficiency in PyTorch or TensorFlow. Computer Vision: Extensive experience with OpenCV, YOLO (v8/v10/v11), and Detectron2. Geometry & Graphics: Experience with 3D reconstruction, point clouds, or mesh generation (e.g., using libraries like Open3D or Trimesh). Data Pipeline: Experience with data labeling tools (CVAT, Roboflow) and handling technical/architectural drawings. Backend: Proficiency in Python and building high-performance APIs for ML inference. Apply tot his job