Start here: What is Aegis?

Plain words, no buzzwords: - Aegis lets you train ML models without pooling raw data in one place. - It protects people’s privacy automatically while training. - It gives you simple controls, clear reports, and production‑ready ops.

Key ideas (kept simple) - Differential Privacy (DP): adds a tiny amount of noise so individual records can’t be picked out. - Federated Learning (FL): each site trains locally; only model updates are shared. - Guardrails: permissions (RBAC), encrypted connections (mTLS), and audit logs.

Is Aegis for me? - You work with sensitive data (healthcare, finance, public sector, education). - You need to collaborate across multiple sites or companies. - You must explain privacy protections to non‑technical stakeholders.

What you get on day one - One command to start a demo with dashboards - Easy privacy settings (pick a privacy level, we do the math) - Built‑in federated training that tolerates slow or unreliable sites - Click‑to‑generate reports for compliance reviews

Quick start (5–10 min) 1) Start locally: see docs/scripts/start_local.sh 2) Open the API docs at http://localhost:8000/docs 3) Start a federated run with a balanced privacy setting 4) Watch Grafana at http://localhost:3000, then export a report

Next steps - New to terms? See basics/glossary.md - Curious about value? See features/overview.md