What is De-Moloch

De-Moloch's purpose is preventing abuse of Big Data. De-Moloch provides an intuitive interface for series of private machine learning libraries and tools, all combined into one, De-moloch offers a way for not only Data providers to secure their data while still allowing ML models to use it, but a way for users to potenially sell their own data to Web 3.0 projects if they choose.

De-Moloch was inspired by Fred Ersham's overview of decentralized machine learning model marketplaces. Specifically, this was in response to his concerns about these markeplaces creating a new "Moloch". I this context, this refers a system in which people make seemingly small sacrifices for short-term gain, but with the result of potentially disastrous results in the long term. One can point to the abuse of centralized data, and the approaching existential threats from AI, as examples of this. It was decided that specialized tools and libraries were not enough to stop these. A community/organization with the sole purpose of making these private machine learning tools available to everyone is needed.

This is where we come in. We're building a decentralized ML marketplace that can work with existing organizations, as well as process more than just financial data.

NOTE: WE ARE NOT DOING A TOKEN SALE. Yes, De-Moloch is a Decentralized App. However, De-Moloch is being designed to allow for integration with a litany of popular cryptocurrencies, as well as fiat currencies.

User-friendly Design

We're building an intuitive interface for people with little CLI experience (or perhaps even none at all) so they can use it on either Windows, Mac, or Linux.

Open Source

Part of DeMoloch's success will come from all materials for the project being open-sourced, allowing security engineers everywhere to spot possible flaws so they can be fixed in time.

Adaptable and customizable

This project is capable of working with many DApp projects, but it is not it's own token. We are not doing a token sale, rather we want to create a tool that all organizations (established, startup, centralized, decentralized) handling large amount of data can use.


There are many open-source commuities and academics working on Private Machine learning. However, if we want to prevent something even worse than the Cambridge Analytica Scandal, we need an organization specifically devoted to helping companies and groups implement private machine learning.

Flexible and Multipurpose

DeMoloch will be able to ingest many types of data schemas, both through the use of "universal schemas" (which can be easily applied to Social Media Data), as well as tools for automatically adapting to unconventional schemas (which can easily be applied to Academic Research Data)

Multiple Security Layers

Multiple components of the stack, from the use of Metamodels, to homomorphic encryption, to the use of federated machine learning, are designed to both protect data from abuse by 3rd parties, as well as reduce the Moral Hazard from centralization

Many labs working on treatments for diseases (like Alzheimer's & Aging) often hav their data siloed away from each other. Even organizations like Human Longevity Inc. or Google Calico might not have enough data. De-Moloch can act as a platform for creating incentives for data sharing, as well as common schemas that could result in commonly-accepted schemas for certain fields (adding some integrity to the data collection itself)

What's often a problem for many distributed DApp projects is keeping track of usage, while keeping the token or DApp decentralized. By using private machine learning tools, analytics can be provided with the user remaining anonymous if needed.

Both startups and large companies struggle with data protection. John Hancock, for example, will likely struggle with user data safety that it collects from fitbits for interactive policies. Private ML eliminates that risk by eliminating much of the need for collecting the data in the first place.

John Hancock recently announced their transition to interactive life incurance policies. This means collecting data from health trackers like Fitbits and Apple watches. An easy-to-understand federated machine learning pipeline could allow John Hancock to keep this policy, while avoiding getting hacked by bad actors who would wish to gain info on locations or keys to financial information.

The Cambridge Analytica scandal is still very fresh. With the use of data ingestion schemas for services like Facebook or Twitter, 3rd parties could gain global insights without being granted unnecessary access to users' entire social media accounts.


Distributed Model Trainings


Ready for Beta testing


Alpha Testers so far


Cups of Coffee consumed


Siraj Raval

Our team originally met on Siraj's Wizards Slack. After Siraj pointed us in the direction of many resources on Distributed AI (and produced some of his own resources). The Project took form. None of this would have been possible without Siraj's help. His Youtube and social media channels have been a steady source of easy-to-digest summaries of how to use and create DApps. In many ways the project is aligns with Siraj's mission of democratizing machine learning. More specifically, we're democratizing private machine learning.

Andrew Trask

Several team members were also part of the OpenMined community, an open-source library for private machine learning created by Andrew Trask. Andrew has made great strides in growing the OpenMined community (which recently celebrated it's 1000th Github star after existing for only 1 year). Part of the Purpose of DeMoloch is to provide resources to developers and researchers and mathematicians devising ways of improving and implementing such libraries (including but not limited to OpenMined), as well as taking care of tasks like helping organizations set up and connect to distributed ML model marketplaces.