3 hours ago
Toronto, Canada or Ottawa, CanadaSenior
Base Salary
$162k - $180k/yr
Responsibilities
- Design and train deep neural network models that perform object detection and tracking using temporal information.
- Evaluate state-of-the-art research papers, implement their concepts, and adapt them into production-grade solutions.
- Customize model architectures, loss functions, internal data flows, and data augmentation strategies.
- Optimize models for accurate, efficient, real-time inference and on-device deployment.
- Develop training recipes and post-training strategies for data-constrained environments.
Requirements
- 5+ years of proficiency with Python and PyTorch.
- 3+ years of proficiency with C++ for production deployment and optimization.
- Deep theoretical and practical understanding of object detectors such as Transformers, YOLO variants, and R-CNNs, as well as tracking algorithms such as DeepSORT, Kalman Filters, and Optical Flow.
- Proven experience modifying model architectures beyond out-of-the-box APIs through extensive experimentation.
- Experience improving model generalization with limited data using transfer learning, domain adaptation, or few-shot learning.
- Strong understanding of linear algebra and probability as applied to custom loss functions and geometric 3D vision.
- Preferred: hands-on experience with 3D point cloud data and LiDAR.
- Preferred: experience with TensorRT, ONNX Runtime, or edge-specific hardware such as NVIDIA Jetson.
Benefits
- Base pay is listed as $162,000-$180,000 and the role may also be eligible for equity and benefits.
- The employer supports an equal-opportunity, diverse, healthy, and safe workplace and provides accommodations for disabilities or special needs.
Categories
ML EngineeringRobotics
