Data Scientist Intern — WiFi Sensing & Human Activity Recognition
- Machine Learning
- Signal Processing
- Time Series
- Edge AI
Zoe Care is building a Physical AI platform that interprets human activity from variations in ambient Wi-Fi signals.
Our technology combines WiFi Sensing, signal processing, machine learning and Edge AI to understand what is happening in a real environment, without cameras or wearables, with a privacy-respecting approach.
Our first application, ZoeFall, detects falls of older adults using a discreet device installed in their living space.
After several real-world pilots, we are entering a new phase: improving the accuracy, robustness and generalization of our models in order to deploy our technology at scale.
The scientific challenge
Recognizing human activity from variations in a Wi-Fi signal alone is a particularly demanding scientific problem. Our technology must:
- acquire and interpret complex radio signals;
- extract relevant information from noisy time series;
- distinguish activities and events that can be very similar;
- remain accurate across different rooms and configurations;
- run locally on resource-constrained devices.
The work sits at the intersection of machine learning, deep learning, signal processing, radio telecommunications, time-series analysis, experimental modeling and embedded artificial intelligence.
Your missions
- Building, qualifying and structuring datasets from experimental campaigns
- Exploring and visualizing complex Wi-Fi data
- Signal preprocessing, filtering and transformation
- Developing human activity classification models
- Designing training, validation and test protocols
- Optimizing precision, recall and false positives
- Improving robustness to changes in rooms, users and configurations
- Analyzing errors and edge cases observed in real-world conditions
- Optimizing models to run on Edge AI devices
- Scientific watch and experimentation with new approaches from research
You may also take part in measurement campaigns and pilot deployments in nursing homes (EHPADs).
Your profile
Engineering student or Master’s (M2) student, for an end-of-studies internship or gap year, specializing in:
- machine learning
- data science
- artificial intelligence
- signal processing
- applied mathematics
- or an equivalent field
Ideally
- Very good Python skills
- PyTorch, TensorFlow, Scikit-learn or equivalent
- Strong probability, statistics, optimization and linear algebra fundamentals
- Experience with time series or sensor data
- Rigorous experimentation and evaluation methodology
- Ability to analyze open-ended problems and formulate hypotheses
- Autonomy and curiosity
Bonus
- Digital signal processing
- Telecommunications / radio signals
- CNN, RNN, LSTM, Transformers
- Anomaly detection / Human Activity Recognition
- Quantization, compression, embedded AI
- Applied research & physical systems
What you’ll find at Zoe Care
- A scientific topic at the frontier between research and product
- Direct impact on a deeptech technology deployed in real-world conditions
- Access to the CentraleSupélec scientific environment
- Direct mentoring by the CTO, PhD in physics and associate professor
- Strong autonomy and ownership of experiments
- Short cycles between data collection, modeling, testing and improvement
- An impact-driven project serving older adults and care professionals
- The entrepreneurial and scientific community of 21st by CentraleSupélec
- Weekly sports activities
- Access to the university restaurant