Machine Learning Engineer – Computer Vision
$100,000–$110,000 year
HybridToronto, Ontario, Canada
Job Summary
Design, train, and fine-tune state-of-the-art deep learning models for image classification and object detection. Optimize and convert TensorFlow models to highly efficient TensorFlow Lite artifacts for edge inference, handling quantization and hardware acceleration. Maintain and enhance MLOps capabilities on Vertex AI Kubeflow Pipelines, architecting deployment flows to save optimized models to Google Cloud Storage and manage versioning. Implement automated evaluation gates to ensure new models outperform production versions before edge deployment. Requires 3-6 years in ML engineering with expertise in CNNs, PyTorch, and Python. Hybrid schedule with two days in Toronto.
Required Qualifications
- 3- 6 years in Machine Learning Engineering, preferably Computer Vision
- Strong mathematical and architectural understanding of deep learning concepts, specifically Convolutional Neural Networks (CNNs) and standard object detection architectures
- Deep, hands-on expertise with TensorFlow 2.x and/or PyTorch
- Proven experience optimizing deep learning models for edge devices using TFLite (e.g., post-training quantization, pruning, handling custom ops)
- Strong proficiency in Google Cloud Platform, specifically building and running custom components in Vertex AI Pipelines (KFP)
- Advanced programming skills in Python, with experience containerizing ML workloads using Docker
- Solid understanding of Google Cloud Storage (GCS) for managing massive datasets and handling model artifact hand-offs
- Critical thinking
- Effective communication skills – verbal and written
- Problem solving
- Dealing with complexity
Desired Qualifications
- Hands-on experience with the Ultralytics YOLOv8 ecosystem, specifically bridging PyTorch YOLO weights to TensorFlow/TFLite edge deployments
- Experience using Google Cloud Composer (Apache Airflow) to schedule and trigger complex ML training pipelines based on data arrival or model drift
- Familiarity with Google Cloud Dataflow (Apache Beam) for large-scale, parallelized image preprocessing, augmentation, and dataset formatting (e.g., generating TFRecords)
- Experience with continuous integration and continuous deployment practices specifically tailored for machine learning models
- Knowledge or experience in Generative AI architectures, with experience building Retrieval-Augmented Generation (RAG) pipelines and developing multi-agent systems
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.