Senior Computer Vision Engineer

Innovasea•Bedford, NS

2d•Hybrid

About The Position

Innovasea is seeking an experienced Computer Vision Engineer (Algorithm Development) to join an energetic team in Canada. This role will focus on shaping the next generation of ocean farming products using underwater perception, 3D vision, and modern deep learning. The successful candidate will lead the design, development, evaluation, and deployment of computer vision algorithms for real-time perception and analytics in challenging underwater environments.

Requirements

Strong foundation (10+ years) in computer vision + deep learning: object detection, instance/semantic segmentation, multi-object tracking
Proven experience building training/evaluation pipelines: dataset management, augmentation strategies, class imbalance handling, metrics design, test set hygiene, regression testing for models, experiment reproducibility and tracking
Proficient in Python for ML development and C++ for production integration
Hands-on experience with modern ML frameworks: PyTorch (preferred)
Experience with deployment/optimization: TensorRT / ONNX, mixed precision, quantization, performance profiling, latency/throughput optimization
Strong software engineering fundamentals: clean architecture, modular code, testing, CI/CD basics
Git-based workflows, code review culture
Excellent communication and collaboration across functions: ability to convert ambiguous problems into actionable experiments and deliverables
Ability to maintain confidentiality and work in a fast-moving environment

Nice To Haves

Experience with underwater vision (turbidity, low-light, motion blur, color shift), domain adaptation, or robustness methods
Experience with active learning, human-in-the-loop labeling, uncertainty estimation, drift detection
Familiarity with MLOps tooling (Weights & Biases, MLflow, DVC, data lineage, model registry)
Exposure to distributed training, multi-GPU, or cloud training workflows

Responsibilities

Design and implement computer vision and deep learning algorithms for underwater applications (detection, tracking, segmentation, pose/key points, 3D reconstruction)
Develop robust underwater image enhancement and domain adaptation approaches (color correction, dehazing, low-light, turbidity robustness)
Create scalable training and evaluation pipelines: dataset curation, labeling workflows, augmentation, model training
Define and track algorithm KPIs (accuracy, precision/recall, tracking stability, 3D error, latency, throughput) and run ablation studies to drive continuous improvement
Optimize models for real-time deployment on edge hardware (Jetson / GPU): quantization, pruning, TensorRT, batching, pipeline profiling
Collaborate with software/hardware teams to integrate models into production systems (APIs, streaming pipelines, monitoring, versioning, rollback strategies)
Work with marine biology and ocean engineering stakeholders to ensure outputs support fish well-being, operational excellence, and scientific validity
Travel domestically/internationally for R&D site work, system validation, and customer visits