Data Science,
Powered by Federated Learning

Our platform leverages federated learning, enabling data scientists to train advanced models without compromising data privacy.

Browser, SDK & CLI

Data scientists can use our SDK library, run in their local environment alongside existing data and models. Results from these jobs can be viewed in a browser through our web application.

Task Runners

Task runners perform operations locally, returning privacy-preserved results to a central orchestrator to produce the final output showing model impact and other analytical results.

Federated Learning Server

The Federated Learning Server orchestrates the communication flow between the different tasks running on the Task Runners, enabling users to perform data science jobs without data moving.
federated data science

As easy as working with centralized data

With tooling for the entire data science workflow, provides a seamless experience from exploration to model training.
Exploratory data analysis: Create visualizations and calculate basic statistics on individual datasets or in aggregate.
Feature engineering: Calculate new features on distributed datasets.
Model training: Configure and train almost any model type, including deep learning models, GLMs, CNNs, LSTMs, Transformers, FFNNs, and decision trees.

Key Principles of Federated Data Science

Federated data science allows data scientists to perform tasks without moving data. Here are some core principles to remember when using federated data science capabilities.

Data Never Moves

Unlike other privacy preserving methods (e.g. clean room, enclave, etc.), in a federated data science system, raw data never moves out of the local environment.

No Individual Data Exposure

In a federated system, a data scientist can never access individual data records from any client. Instead, they can only train models and perform analyses with full privacy-protection of the original data.

Comprehensive Data Governance

The data owner or custodian can decide what data can be used for which tasks at what time. They are able to grant or revoke permission to any dataset at any time.
How it works

Get Started in 3 Simple Steps

Connect Your Data

Setup Task Runners on your cloud platform(s) to connect to the Federated Learning Server

Register Your Data

Prepare and register dataset metadata using your established Task Runner


Via the SDK, securely run data science jobs with internal and third party data through the Federated Learning Server

Frequently asked questions
Contact Us
What is a Task Runner and how is it deployed?
What is the Federated Learning Server?
Is data ever moved or copied?
How is data protected both in-transit and at rest?
Can, or any of its dependencies see any of my data?