Machine Learning Engineer

Machine Learning Engineer (2D→3D Reconstruction & Workflow Intelligence) OpEase Technologies builds a high-precision, web-based surgical planning platform for orthopedic and spine surgeons. Doctors use OpEase to securely store patient data, upload X-rays, calibrate, measure, and plan surgeries through advanced geometry tools and clinical logic. Role Overview: We’re hiring an ML Engineer who will own the core AI systems powering OpEase — specifically: 1. Reconstructing 3D anatomical structures from orthogonal 2D X-rays, and 2. Building intelligent auto-selection and auto-suggestion logic for measurement and planning tools inside our surgical workflow. This is not a research-only role. You will be responsible for designing, training, validating, and deploying production-grade ML systems that directly impact surgical decision-making. Clear problem statements and datasets will be provided; you are expected to execute with speed and rigor. What You Will Build (Very Specific): - A robust 2D→3D spine/long-bone reconstruction model using dual-view X-rays (AP + lateral) - Landmark/keypoint detection models for vertebrae, femur/tibia, pelvis, etc. - Heatmap regression networks for anatomical feature extraction - A model-driven auto-selection system that identifies which OpEase tool the surgeon requires based on image context and user behaviour - End-to-end inference pipeline integrated into our MERN + Cornerstone-based viewer - Continuous evaluation pipelines for accuracy, latency, and failure-case analysis Responsibilities: - Architect and train models for 2D→3D anatomical prediction using multi-view geometry, implicit fields, NeRF/DVGO variants, or transformer-based approaches - Build landmark detection modules for calibration, templating, and surgical planning - Design the autosuggestion engine: tool intent prediction, context modelling, clinical-rule integration - Manage data pipelines for X-ray preprocessing, augmentation, versioning, annotation QC, and synthetic dataset generation - Validate models with surgeons; refine based on clinical feedback - Deploy models to production (REST endpoints, ONNX/TensorRT optimization, GPU/CPU fallback) - Maintain experiment logs, metrics dashboards, and detailed model documentation Requirements (High Priority & Non-Negotiable): - Minimum 4 years of full-time experience in ML/Deep Learning with shipped models in production - Strong experience in computer vision for geometry problems: keypoints, reconstruction, pose estimation, volumetric prediction - Hands-on expertise with PyTorch, multi-GPU training, and advanced optimization techniques - Prior work with DICOM/X-ray/medical imaging OR demonstrably adjacent experience (e.G., industrial CV, robotics perception, pose estimation) - Proven ability to independently take a model from idea → dataset → training → evaluation → production - Strong mathematical grounding in 3D geometry, camera models, coordinate transforms, and projection systems - Excellent documentation and communication skills Bonus (Big Plus): - Experience with NeRFs, implicit neural representations, depth inference, or differentiable rendering - Experience building autosuggestion systems, ranking models, or intent prediction in complex workflows Why Join: - You will own the foundational AI layer for India’s most advanced orthopedic planning platform - Clear, well-scoped problems and direct access to clinicians who use your models - Chance to build category-defining medical AI from the ground up - High ownership, high-impact role in a company scaling rapidly across India and global markets Hybrid role with periodic clinical onsite work. Compensation: ₹18–24 LPA + ESOPs.