Working Student Machine Learning

What you'll do

• Design and prototype ML models tailored for AR use cases under embedded hardware constraints.
• Set up datasets and baselines for AR tasks like detection, tracking, and segmentation.
• Implement and train models using PyTorch including data pipelines and evaluation scripts.
• Explore efficiency techniques such as pruning, quantization, and sparsity to optimize performance.
• Profile models on edge accelerators to analyze FLOPs, latency, memory footprint, and bandwidth.
• Conduct ablation studies to evaluate trade-offs between accuracy, latency, and energy consumption.
• Develop a reproducible codebase with pre-trained checkpoints for the final thesis project.
• Produce a high-quality thesis report and optional paper-style write-up of research findings.

What we're looking for

• Must be currently enrolled in a Master’s program in Computer Science, Engineering, AI, Robotics, or a related field.
• Degree program must allow for a Master's thesis or graduation project with an external organization.
• Strong background in linear algebra, probability, and optimization is required.
• Deep learning fundamentals including backpropagation, regularization, and basic model architectures are required.
• Must have hands-on experience training deep learning models for computer vision using PyTorch or a similar framework.
• Proficiency in Python and standard ML tools like NumPy, Git, and experiment management is required.
• Preferred experience with event-based/streaming vision, model compression (pruning, quantization), or efficient architectures for embedded systems.
• Familiarity with on-device ML toolchains like TensorFlow Lite or ONNX Runtime is preferred.