Data Science Architect- Computer Vision

We are looking for a Data Science Architect (Computer Vision) to lead the design and evolution of our data science systems across machine and cloud platforms. As a key member of the R&D team, you will shape the architecture that powers deep learning inference on TOMRA Food sorters and drive innovation in our cloud-based MLOps platform that was developed to train the models. This role blends technical leadership, system design, and strategic foresight to deliver scalable, high-performance solutions that create real customer value.

Role and Responsibilities:

System Architecture & MLOps:

• Design vision models for real-world food sorting applications
• Design architecture for real-time deep learning inference on TOMRA sorters, including GPU specifications. The tasks include object detection, segmentation and more.
• Design DL vision models (CNN-based like ResNet, MobileNet, EfficientNet & Transformed-based) by selecting a suitable backbone, neck and head and testing the performance of those combinations. Innovations in these areas are highly welcomed.
• Maintain & improve architecture of in-house cloud-based MLOps platform (data science pipelines).
• Design & advise on new features for the MLOps platform.
• Follow latest improvements in AzureML ecosystem and come up with efficiency improvements.
• Optimise model lifecycle process and come up with performance improvements for our DL models in order to create customer value.
• Produce architectural plans for our data science solutions and ensure that these are well communicated and understood by the rest of the data science team.

Technical Leadership:

• Provide technical leadership and guidance to data scientists.
• Stay up to date of industry trends and emerging technologies to ensure our systems remain cutting-edge.
• Understand customer needs and assist in translating these to technical product requirements.
• Follow latest DL model trends and push improvements to the data science team (this includes making use of new libraries).