Working Student - Machine Learning

TL;DR

Working Student - Machine Learning (AR): Exploring efficient on-device ML for AR glasses using event-based sensing and embedded processors with an accent on model architecture, efficiency techniques, and hardware constraints. Focus on designing accurate ultra-efficient models for real-time AR tasks, evaluating trade-offs in accuracy, latency, memory, and energy, and delivering proof-of-concepts on AR hardware like Spectacles.

Location: Eindhoven, the Netherlands, with a minimum of 4 days per week in the office

Company

Snap Inc. is a technology company building hirify.global visual messaging app, Lens Studio AR platform, and Spectacles AR glasses.

What you will do

• Define and drive a research direction in efficient on-device ML for AR, emphasizing event-driven or embedded processors.
• Design and prototype ML models for AR tasks under embedded constraints, such as event-based vision models or lightweight CNNs/Vision Transformers.
• Set up datasets, baselines, and evaluation metrics for AR tasks like detection, tracking, segmentation, and gesture recognition.
• Implement and train models in PyTorch, exploring efficiency techniques like sparsity, pruning, and quantization.
• Profile models on embedded-like conditions and communicate findings through ablation studies, thesis report, and reproducible codebase.
• Demonstrate proof-of-concepts on AR hardware showcasing improvements in performance and efficiency.

Requirements

• Currently enrolled in a Master’s program (Computer Science, Electrical/Computer Engineering, AI, Robotics or related) with thesis/graduation project allowed in collaboration with external organization
• Strong background in linear algebra, probability, optimization, and deep learning fundamentals.
• Hands-on experience training deep learning models for computer vision with PyTorch (preferred), CNNs, and/or vision transformers.
• Proficiency in Python, NumPy, Git, and basic ML tooling.
• Interest in reproducible codebases.

Nice to have

• Experience with event-based vision, model compression (pruning, quantization, distillation), or efficient architectures for embedded applications.
• Familiarity with embedded ML toolchains like TensorFlow Lite or ONNX Runtime.
• Exposure to performance profiling and systems concepts (FLOPs, latency, memory).
• Experience with AI-assisted development tools.

Culture & Benefits

• "Default Together" policy expecting 4+ days per week in office to build culture and collaboration.
• Comprehensive benefits including paid parental leave, medical coverage, mental health support, and compensation packages sharing long-term success.
• Equal opportunity employer committed to diversity and inclusion.