Scientific Assistant – Real-Time Generative Models / Computer Vision

Implement, optimize, and deploy deep-learning image-enhancement models for real-time operation. Translate existing and new architectures into efficient real-time processing pipelines connected directly to an industrial camera system. Profile and optimize models for low-latency inference; implement and adapt neural-network architectures with real-time constraints in mind; integrate camera acquisition and preprocessing pipelines; and contribute to project deliverables and scientific manuscript preparation.

The Biomedical Image Computing Group develops algorithmic solutions to enable automatic interpretation and forming of biomedical images, aiming to increase efficiency and accuracy in clinical practice and research. The successful candidate will join the BMIC team at ETH Zurich and also be part of the Computer Vision Lab (CVL). The project is conducted in collaboration with an industrial implementation partner, who will provide support with hardware and software integration.

ETH Zurich is one of the world’s leading universities specialising in science and technology. We are renowned for our excellent education, cutting-edge fundamental research and direct transfer of new knowledge into society. Over 30,000 people from more than 120 countries find our university to be a place that promotes independent thinking and an environment that inspires excellence. Located in the heart of Europe, yet forging connections all over the world, we work together to develop solutions for the global challenges of today and tomorrow.

• MSc degree in computer science, machine learning, medical image computing, or a related field.

• Familiarity with neural-network-based image processing pipelines.

• Excellent programming skills and experience with Python.

• Experience implementing and validating neural-network architectures using PyTorch.

• Demonstrable experience with model deployment/acceleration (PyTorch, ONNX, TensorRT, CUDA/C++, etc.), profiling tools, and camera/sensor integration is a plus.