# Machine Learning / Software Engineering Thesis Student at Snap Inc
## Project Overview
Join Snap Inc's team to explore efficient on-device machine learning for augmented reality applications on embedded and event-driven processors. This thesis project focuses on combining modern deep learning with event-based sensing to enable always-on, real-time AR experiences within strict latency, energy, and bandwidth constraints.
## What You'll Do
- Design and prototype ML models tailored to AR use cases under embedded constraints (event-based vision models, lightweight CNNs/Vision Transformers, hybrid frame+event pipelines)
- Set up datasets and baselines relevant to AR tasks (detection, tracking, segmentation, gesture/interaction) with evaluation metrics across accuracy, latency, memory usage, and energy
- Implement and train models in PyTorch with reproducible data pipelines, training loops, and evaluation scripts
- Explore efficiency techniques such as sparsity, pruning, quantization (PTQ/QAT), and event-based representations
- Profile models under embedded-like conditions using simulators, profiling tools, or edge accelerators
- Communicate findings through ablation studies, thesis report, and reproducible codebase with pre-trained checkpoints
## Expected Outcomes
- Proof-of-concepts on AR hardware (e.g., Spectacles) showcasing real-world impact
- Measurable improvements in runtime performance, efficiency, and adaptability for representative AR tasks
- Insights into model–system co-design for low-power, on-device ML
- Contributions to ML frameworks and deployment strategies for embedded AR systems
- High-quality thesis report with reproducible code and results
## Requirements
- Currently enrolled in a Master's program (Computer Science, Electrical/Computer Engineering, Artificial Intelligence, Robotics, or related field)
- Degree program allows Master's thesis/graduation project with external organization
- Strong background in linear algebra, probability, optimization, and deep learning fundamentals
- Hands-on experience training deep learning models for computer vision
- Proficiency in Python and standard ML tooling (NumPy, PyTorch, Git, experiment management)
## Preferred Experience
- Event-based or streaming vision
- Model compression techniques (pruning, sparsity, quantization, knowledge distillation)
- Efficient architectures for embedded/real-time applications
- Embedded/on-device ML toolchains (TensorFlow Lite, ONNX Runtime)
- Performance profiling and systems concepts