About the role

Développeur Vision/IA senior développant des modèles de vision par ordinateur pour le tri robotique industriel de Waste Robotics. Déploiement MLOps temps réel et optimisation du picking haute vitesse.

Responsibilities

Train, evaluate, compare, and deliver detection and segmentation models (YOLO, DETR, Mask R-CNN)
Diagnose errors and report comprehensive metrics (precision, recall, F1, AP/mAP, latency)
Deploy real-time inference at the edge
Conduct parity testing across environments
Optimize latency and throughput
Ensure traceable and reversible releases
Develop evaluation protocols robust to variations in conveyors, lighting, and cameras
Determine the minimum dataset required for a new site
Prioritize the most informative images for active annotation
Monitor the quality of model pre-annotations
Specify and validate RGB-D cameras, optics, and commissioning conditions
Advance picking performance at flagship facilities to 95%, with transparent class-level reporting
Collaborate with multidisciplinary teams in AI, software, hardware, operations, and domain expertise

Senior ownership of production Python systems, including design, testing, deployment, and support
Hands-on experience with object detection or instance segmentation, applying evaluation rigor beyond a single metric
Proven MLOps practices and deployment with PyTorch, ONNX, TensorRT, or equivalent technologies
Experience with domain shift, imbalanced data, or active learning
Strong software engineering skills, including modular design, automated testing, and documentation
Nice to have: industrial imaging (RGB-D, calibration, challenging environments), CUDA/GPU profiling, cloud-based data pipelines, and model-tracking platforms
Excellent communication and collaboration skills within a multidisciplinary team spanning AI, software, hardware, operations, and domain experts
Ability to mentor peers and communicate technical trade-offs directly, particularly during code and design reviews
Self-directed in organizing your work, with sound judgment in managing priorities
Analytical mindset, with the ability to challenge incomparable results and clearly explain experimental limitations
Professional communication skills in French and English
Currently residing in the Montreal or Trois-Rivières area; remote work from outside these regions and an intention to relocate in the future do not meet this requirement

4 weeks of vacation
A day off on your birthday
Comprehensive insurance coverage, including dental, an employee assistance program, and telemedicine
Team activities
End-to-end visibility into robotic picking performance, from the sensor to the production robot
A tangible impact through technology deployed across multiple countries in support of the circular economy