Senior ML Engineer — Geospatial Vision for Infrastructure

Citylogix; Montreal (administrative region); On-site; CAD 120,000 - 170,000; Full time

Job summary

Citylogix in Montreal, a data and analytics company, is seeking a Senior Machine Learning Engineer to research, develop, and deploy transformer-based segmentation models for infrastructure defect detection and road asset mapping from large-scale imagery.

You will own models end-to-end, improving accuracy and generalization across camera systems, and collaborate with DevOps to scale training and inference on AWS/SageMaker, while maintaining reproducible pipelines and clear documentation.

Qualifications

• Minimum 5 years of applied ML/CV experience, or MSc/PhD plus 3+ years
• Strong hands-on PyTorch proficiency
• Experience with transformer-based segmentation on real-world image data
• Experience with cloud deployment on AWS/SageMaker
• Strong Python engineering skills and software practices
• Excellent communication and collaboration skills

Responsibilities

• Research, design, train, and evaluate deep-learning models for semantic, instance, and panoptic segmentation in infrastructure contexts
• Fine-tune transformer-based segmentation architectures and improve domain generalization across camera systems
• Consolidate multi-task checkpoints into unified production models
• Collaborate with DevOps to train, deploy, and monitor models on AWS/SageMaker
• Build reproducible training pipelines and evaluation benchmarks
• Diagnose model failures and data-quality issues, driving measurable improvements
• Maintain model documentation and versioning standards

Skills

PyTorch

Deep learning

Transformers

Domain generalization

Hugging Face

Detectron2

AWS SageMaker

Python

Communication

Education

Master's or PhD in relevant field

Tools

Mask2Former

OneFormer

DETR

SageMaker

Job description

Citylogix in Montreal, a data and analytics company, is seeking a Senior Machine Learning Engineer to research, develop, and deploy transformer-based segmentation models for infrastructure defect detection and road asset mapping from large-scale imagery.

You will own models end-to-end, improving accuracy and generalization across camera systems, and collaborate with DevOps to scale training and inference on AWS/SageMaker, while maintaining reproducible pipelines and clear documentation.