Autonomous cloud storage environment that protects against both contemporary threats and future quantum cryptanalysis.
- Hybrid encryption: AES-256-GCM for data + CRYSTALS-Kyber (ML-KEM-1024) for key encapsulation — NIST PQC standard
- AI threat detection: XGBoost model trained on 10,000 synthetic attack logs, monitoring 20 behavioral vectors including geo-velocity, API spike rates, and rolling byte transfers
- Autonomous self-healing: When anomaly score > 0.90, system automatically severs connection, isolates account, rotates all keys — within 8 seconds. No human intervention required.
- Harvest Now Decrypt Later (HNDL) protection: Exfiltrated ciphertext becomes mathematically obsolete after key rotation
- MinIO object storage
- RabbitMQ event-driven architecture
- XGBoost risk engine with SMOTE class balancing
- Python-based cryptographic worker for key rotation
- Docker containerized deployment
| Component | Algorithm |
|---|---|
| Bulk encryption | AES-256-GCM |
| Key encapsulation | CRYSTALS-Kyber / ML-KEM-1024 |
| Threat model | Shor's Algorithm resistant |
Final year capstone project — MS Ramaiah University of Applied Sciences, Department of CSE, April 2026.
- Adhithyan S
- Hrishikesh E
- Kartik Nair
- Mohammed Sahis NP
Supervisor: Ms. S. Maragatham