Computer Vision Research Scientist Job Description Template

The computer vision research scientist advances methods for understanding images and video, working on problems such as detection, segmentation, tracking and generation. The field moves quickly, so the role requires both depth in the fundamentals and the ability to evaluate new work critically rather than adopting it because it is recent.

Typical Duties and Responsibilities

• Conduct research on computer vision methods and architectures
• Design experiments with appropriate datasets and baselines
• Implement and reproduce published methods for comparison
• Develop novel approaches to open problems in the domain
• Evaluate models against realistic rather than only benchmark conditions
• Publish results and present at conferences
• Collaborate with engineers on productionising research
• Curate and analyze datasets including their biases
• Mentor junior researchers
• Track and assess developments in the field

Education

• PhD in computer vision, machine learning or a closely related field

Required Skills and Experience

• PhD with a computer vision publication record
• Deep knowledge of vision architectures and training methods
• Strong Python and PyTorch or an equivalent framework
• Experience with large scale image or video datasets
• Rigorous experimental methodology
• Awareness of dataset bias and generalisation failure
• Clear scientific writing and presentation
• Ability to assess new work critically

Preferred Qualifications

• Publications at CVPR, ICCV, ECCV or comparable venues
• Experience with multimodal or generative vision models