Fully Homomorphic Encryption Studio IDE

Homomorphic encryption is a cryptographic technique that allows you to perform operations on encrypted data without needing to decrypt it first.

Comment

Inspiration

Imagine a world where your data transactions are all public as they are secure, preserving privacy. I was thinking of an IDE bringing a user-friendly experience where you can build programs running on encrypted data, while keeping your sensitive information encrypted. Say goodbye to privacy concerns and keep your data public!

What it does

Cloud IDE where the user can build fully homomorphic programs (circuits), debug it and share it with others

How we built it

Python - based circuits, python sandbox for code lockbox, Angular front-end, OAUTH2 authentication using .well-known endpoints (currently with Google OAUTH2 )

Challenges we ran into

  • reading tons of academic papers on FHE

Accomplishments that we're proud of

Deployed fhe-studio.com in the cloud and pushed the docker image. https://hub.docker.com/r/alextmn/fhe-studio

What we learned

  • FHE is very difficult to learn and its computationally extensive, so it needs all kind of tricks (data structures and other methods to speed it up).
  • FHE natively supports addition and multiplication only, but other functions (sigmoid, ReLU) could be implemented quite easily as well.
  • FHE support the bootstrapping, very cool method, a type of reencryption, allowing to build very deep computations, e.g. encrypted large language models (LLM)
  • FHEs are Post-Quantum encryption schemas (Lattice)
  • Lots of fun!

What's next for Fully Homomorphic Encryption Studio IDE

  • Employ Quantization for FHE circuits for matrix multiplication, to make neural networks and other ML models handier
  • Develop WebAssembly FHE Vault

Built With

Share this project:

Updates