Computer Vision Engineer (80-100%)
Posted
Description
- Taking responsibility for modules of our camera tracking and calibration framework.
- Exploring improvements to tracking and calibration in diverse broadcast scenarios.
- Designing and improving algorithms and models for video understanding.
- Prototyping and validating state-of-the-art approaches from AI/ML research papers.
- Building, extending and curating our machine learning datasets.
- Training and deploying neural networks for our real-time C++ pipeline.
- Prototyping internal tools, designing evaluation benchmarks and writing tests.
- Identifying bottlenecks and optimizing performance.
- Working with the testing and operations teams on usability and reliability.
- MSc/PhD in computer vision, computer science, robotics or a related technical field.
- Experience with visual SLAM, visual odometry or camera calibration.
- Experience with deep learning: data pipelines, model architecture and loss design.
- Good knowledge and experience in C++ and Python.
- Good problem-solving and communication skills.
- Experience with multi-view geometry, image matching or optical flow is a plus.
- Experience with foundation models or knowledge distillation techniques is a plus.
- Know-how in real-time GPU computing (CUDA, TensorRT) is a plus.
- Familiarity with testing frameworks and continuous integration is a plus.
- Interest in agentic coding tools is a plus.
- Whether you are starting out or already experienced, we welcome your application.
- A dynamic and fast-growing environment with many possibilities for further development.
- A highly motivated, interdisciplinary, multi-cultural team with focus and vision.
- Several (larger and smaller) team events per year
- A competitive salary and participation in uniqFEED's Phantom Stock Option Plan
- Financial support for training and further education
- Five weeks of vacation per year
- Flexible working hours, home office by prior agreement, workation, and an appealing work environment in our HQ in Glattbrugg/ ZH.
Skills
Sign in to read and join the discussion.