Lilly Tech Systems
Learn Homomorphic Encryption for AI
Master computation on encrypted data. From FHE fundamentals and the CKKS scheme to encrypted ML inference with Microsoft SEAL, Concrete ML, and TenSEAL — build AI systems that never see plaintext data.
Your Learning Path
Follow these lessons in order to build a complete understanding of homomorphic encryption for AI.
1. Introduction
What is homomorphic encryption, why compute on encrypted data, the HE landscape, and real-world applications.
2. HE Fundamentals
PHE, SHE, and FHE schemes. The CKKS scheme for approximate arithmetic, BFV for exact arithmetic, noise management, and bootstrapping.
3. ML on Encrypted Data
Encrypted inference for neural networks, logistic regression, decision trees, and the challenges of non-linear activations.
4. Libraries & Tools
Microsoft SEAL, Concrete ML, TenSEAL, OpenFHE, and HElib. Hands-on examples with each library.
5. Performance
Ciphertext packing, SIMD operations, parameter selection, hardware acceleration, and benchmarking encrypted ML.
6. Best Practices
When to use HE vs other PETs, model design for FHE, security parameters, and production deployment guidance.
What You'll Learn
By the end of this course, you'll be able to:
Understand HE Theory
Grasp FHE schemes (CKKS, BFV), noise budgets, bootstrapping, and parameter selection.
Run Encrypted Inference
Deploy ML models that compute predictions on encrypted data without ever seeing plaintext.
Use HE Libraries
Build encrypted ML applications with SEAL, Concrete ML, TenSEAL, and OpenFHE.
Optimize Performance
Apply ciphertext packing, SIMD batching, and model design techniques for practical performance.