Zama: Homomorphic Encryption Tools

Securing against data breaches and information tracking in cloud services while reading smart contracts is now achievable with tools for fully homomorphic encryption. Zama is one of the crypto companies providing products in this field. 

Rand Hindi, an entrepreneur and deeptech investor, is the founder and CEO of the project. His portfolio encompasses over 50 companies specializing in areas such as privacy, AI, blockchain, and healthcare. 

Rand, a programming prodigy since age 10, initiated his first venture in social network platforms at 14, and by 21, had secured a Ph.D. in bioinformatics. Alongside Zama, Hindi established an AI startup, Snips, eventually acquired by Sonos.

What is Homomorphic Encryption?

Homomorphic encryption is a specialized cryptographic technique that enables the execution of mathematical calculations on obscured information, producing an encrypted outcome. When decrypted, this outcome aligns with the original input data. 

Fully Homomorphic Encryption (FHE), a further refined cryptographic system conceptualized in 2009 by Craig Gentry, is regarded as one of cryptography’s most groundbreaking advancements.

FHE permits manipulations on encrypted data similarly to unencrypted data, eliminating the need for decryption. This facilitates and secures the management of vital data on cloud services while maintaining the confidentiality of the content from the cloud service provider. 

By leveraging FHE tools, developers can build applications that comprehensively preserve user privacy and shield data from unauthorized access and theft, even when server integrity is compromised. The realm of fully homomorphic encryption is currently in its embryonic stage of practical implementation. 

What products does Zama offer?

TFHE-rs

This is the latest iteration of an open-source developer's library, allowing the crafting of services employing Fully Homomorphic Encryption (FHE) in the Rust programming language. It performs logical and integer arithmetic operations on concealed data and is grounded in Learning With Errors (LWE) machine learning.

How FHE is employed within the TFHE-rs library:

• Both a client and a server key are generated.
• The client key, vital for encrypting and decrypting data, needs to be securely stored and kept confidential.
• The server key is essential for conducting operations with encrypted data and is accessible to the public.
• Unencrypted data is encoded using the client key.
• Once encrypted, data can be reverted to its original, unencrypted state using the client key. 

Сoncrete

This accessible program was designed to translate applications, written in Python, into equivalents employing Fully Homomorphic Encryption (FHE). This is essential for the streamlined deployment of FHE, considering it is a complex process. Concrete is built on the LLVM infrastructure, aimed at compiler creation, making it more approachable for developers. 

Сoncrete ML

This software environment allows for the automatic conversion of artificial intelligence applications into their FHE-utilizing counterparts without requiring knowledge in cryptography. Concrete ML includes APIs from the scikit-learn library and the PyTorch framework, elements familiar to developers in the field of machine learning. 

fhEVM

fhEVM is a tool specialized in creating secure and confidential smart contracts on the Ethereum Virtual Machine using homomorphic encryption. It encompasses the JavaScript library fhevmjs, enabling developers to engage with the blockchain in encrypted conditions, thus incorporating FHE in Web3. Additionally, the program features a Solidity library to handle smart contracts utilizing homomorphic encryption effectively. 

On July 21, 2023, the Zama team organized a hackathon at the Ethereum Community Conference to showcase their new product, fhEVM. Attendees had the exclusive opportunity to experience deploying private smart contracts based on homomorphic encryption first-hand. Zama proudly served as an official sponsor of EthCC.