Senior Researcher – Cryptography

Our mission at Zama is to protect people’s privacy by preventing data breaches and surveillance.

Following a recent breakthrough in homomorphic encryption, we are now building a deep learning framework that enables fast and accurate inference over encrypted data, 1000x faster than currently published methods.

We believe privacy-enabling technologies should benefit the widest possible community of developers and security researchers, which is why everything we create will be published and open-sourced.

Zama is founded by Dr Rand Hindi, an serial AI entrepreneur, and Dr Pascal Paillier, who invented some of the most widely used homomorphic encryption schemes. 


Zama is tackling the problem space around the rampant reality of data breaches and surveillance. Currently, companies have no way to protect against unauthorized access to their user data, such as when hackers steal the data or governments subpoenas them. The massive financial losses incurred cannot be prevented with traditional IT security anymore, as hackers have an incentive to hack a company that is proportional to their userbase. This is why the market for data security is growing 20% YoY.

What is being proposed is to use homomorphic encryption to keep the data encrypted server-side while it is being processed. From the user’s perspective, it’s the exact same service. However, the company no longer has access to the data, and so it cannot be stolen. This solves both privacy for users and security for companies.

Conversations have shown demand from companies offering machine-learning services in the cloud, such as facial recognition or speech APIs, as they currently have no way to protect against motivated hackers and governments. By implementing Zama, MLaaS companies would now be able to offer their service encrypted, thereby no longer having to worry about data breaches, localization or privacy.


Our machine learning solution is composed of 3 central elements: 

  • a cryptographic SDK that handles all the cryptographic operations
  • a homomorphic compiler that converts a trained neural network model into an encrypted equivalent
  • a deep learning inference engine that runs the model generated by the compiler, to produce encrypted output from encrypted inputs

Open source by design, we are building the product in a way that maximizes adoption by developers, while making our framework so easy to integrate that no cryptography knowledge would be required to implement it in production.


Your team (and thus you) will be responsible for:

  • discovering new cryptographic techniques to compute on encrypted data
  • working with the engineering and product teams to implement your research into our products
  • design robust benchmarks to test your research and its implementation
  • review the latest published research, and inform the team on potential new applications or changes to our approach
  • work with the entire team to define the research and product roadmaps
  • publishing papers, filing patents and presenting your work at academic conferences


  • PhD in cryptography or equivalent
  • deep knowledge of homomorphic encryption
  • optionally knowledge of machine learning
  • be based in or willing to relocate to Paris, France
  • passionate about privacy and open science