ML Security Guides
Step-by-step guides to securing machine learning systems. Interactive demos, deployment checklists, and hands-on learning.
Adversarial Attack & Defense Demo
See how small input perturbations fool ML classifiers. Visualize FGSM, PGD attacks and defense techniques including input preprocessing and noise filtering.
Federated Learning Introduction
Learn federated learning through an interactive simulator. Visualize N clients training locally, gradient aggregation, and global model updates with privacy budgets.
Secure ML Deployment Guide
Complete deployment security checklist for production ML systems. Container security, API gateway, model serving, secrets management, and monitoring.
Prompt Injection Defense Guide — 12 Techniques That Work
Interactive attack/defense simulator for LLM applications. Test direct injection, indirect injection, jailbreaks, and extraction attacks against 7 countermeasures with traffic-light risk ratings.
How to Choose an Epsilon for Differential Privacy
Fix delta first, compute the noise each epsilon demands, count composition across queries, and record the budget before release.
Pickle Deserialization Risk in Model Files
Loading a pickle runs code. Inspect the opcodes with pickletools, prefer safetensors, and isolate any loader you cannot verify.