Saltar al contenido principal

Curso completo próximamente

Securing Data in Use: Confidential Computing for AI Systems

7 capítulos · 34 lecciones

  1. 1. Data Protection States and the In-Use Gap

    4 lecciones

    Understand why traditional encryption at rest and in transit leaves data vulnerable during processing, and how confidential computing addresses this gap.

  2. 2. Trusted Execution Environments and Hardware Primitives

    5 lecciones

    Learn how TEEs isolate workloads using hardware-enforced memory encryption and access controls across different processor architectures.

  3. 3. Remote Attestation and Cryptographic Verification

    5 lecciones

    Implement attestation protocols that let clients verify workload integrity before releasing sensitive data to protected environments.

  4. 4. Securing RAG Systems and Vector Databases

    5 lecciones

    Design retrieval-augmented generation pipelines that protect confidential documents and embeddings during indexing, search, and context assembly.

  5. 5. Confidential Inference and Model Serving

    4 lecciones

    Deploy inference workloads that protect model weights, input prompts, and intermediate activations from infrastructure operators and co-tenants.

  6. 6. Fine-Tuning and Training with Sensitive Data

    5 lecciones

    Architect training and fine-tuning pipelines that keep proprietary datasets and updated model parameters inside protected memory.

  7. 7. Defense in Depth and Operational Hardening

    6 lecciones

    Combine confidential computing with application security, side-channel mitigations, and supply chain controls for robust AI system protection.

Recibe un correo cuando abra este curso