Cybersecurity · intermediate
Behavioral Biometrics Engine for Continuous Authentication
A security system that authenticates users based on typing rhythm and mouse movement patterns — not static passwords.
Project overview
With the death of static passwords, employers are desperate for experts in Zero Trust and continuous authentication. This project showcases your ability to implement "silent" security layers that stop account takeovers even when credentials have been compromised.
Real-world use case
An e-commerce platform adds this behavioral layer to its checkout process to silently detect and block bot-driven account takeovers and fraudulent purchases without adding friction for legitimate users.
Tools and technologies
- Python
- Scikit-learn
- Flask
- JavaScript
Skills you can demonstrate
Anomaly Detection, Time-series Analysis, IAM
Build roadmap
- Build the data capture frontend
Create a simple login page in HTML/JS that captures keystroke dwell time, flight time, and mouse trajectories. - Design the event schema
Store events in a clean time-series format (JSON schema in the AI prompt below). Think about GDPR early. - Train the baseline model
Use scikit-learn's Isolation Forest on a single user's "normal" typing pattern to detect anomalies. - Build the Flask backend
Create endpoints to /enroll (collect training data) and /verify (score a new session against the profile). - Create a live demo page
Let users type a paragraph to train the model, then try to "impersonate" by typing differently. Show the anomaly score in real time. - Upgrade to LSTM (bonus)
Once the baseline works, replace Isolation Forest with a small LSTM to catch evolving typing patterns over time.
Ways to extend the project
Incorporate an LSTM neural network to improve the system's ability to recognize evolving typing patterns over time for individual users.
How to showcase your work
- GitHub: Include a live demo where users "train" the system on their typing and try to "hack" it.
- LinkedIn: Highlight the anomaly detection accuracy.
- Resume: Focus on Zero Trust architecture and continuous authentication.