WorkPi - OceanDAO Round 9

Part 1 - Proposal Submission

Name of Project

Work X - Compute-To-SSI

Proposal in one sentence

Use Compute-To-Data algorithms on data enclosed in personal Self-Sovereign Identity (SSI) wallets without compromising privacy, we call this Compute-To-SSI.

Description of the project and what problem is it solving

While personal data being protected and governed with SSI wallets can guarantee privacy of that data, currently A.I. algorithms are not able to learn from it. This is a large barrier for adoption. Furthermore we find that the companies that do want to share data using Self Sovereign Identity, prefer to keep their own copy. In some cases like governments this is a necessity, however in many situations it is not.

If private personal data would only exist in SSI wallets, using Compute-To-SSI would enable companies not only to benefit from reduced GDPR risk, it would allow them to learn from this personal data without having to possess it.

Historically, enterprises are very hesitant to share data with other companies, due to privacy concerns and ultimately losing their competitive advantage. This cripples their ability to train an accurate A.I. and gather valuable insights. Besides machine learning algorithms still being able to learn from the data given back to the individual, it could also learn from the data provided by other companies/entities. A Compute-To-SSI mechanism could enable a data driven revolution, as it would grant access to an enormous decentralized pool of data to learn from. By doing so, this will greatly increase the accuracy of these algorithms.

We feel this research area has the potential to let people and organisations benefit from Artificial Intelligence models, which are now exclusively available to tech giants, while preserving privacy.

At WorkPi we are building a decentralized Internet of Jobs, and provide a SaaS solution to enterprises to assess and educate their employees. By doing so, we are creating one of the many valuable use cases that will be enabled by the compute-to-SSI model. After all, besides skills & education information, there are many other types of data that a person can add to their personal data wallet (e.g. health, wearable, cookies, identity or finance data).

This solution will feature the ability to export and import your career related profile to an SSI wallet. In order to become compliant and future proof we are participating in the eSSIF* program of the European Commission, and integrate with the EBSI blockchain**


We strongly believe in giving back personal data to individuals. We see this as a win-win for both the individual as well as enterprises and the key to both fair economic and personal success. However, at this point the problem remains how to learn from this private data.

We decided to work with Ocean Protocol as they have presented innovative solutions, with a strong focus on retaining privacy of data in large silo’s (e.g. hospitals). We aim to take this one step further; directly into the pockets of the people. People will be able to easily give access to specific data as well as revoke it. At the same time, A.I. algorithms will get the ability to learn from this data, all while preserving privacy.

While this will be no easy feat to accomplish, all technical components required to achieve this goal already exist. At WorkPi we are prepared to embark on this grand voyage of the world’s Ocean with you, to improve the ecosystem.

In order to succeed, this project has to be planned carefully and architected precisely with secured partners in the ecosystem.

We want to be very clear about the fact that from the beginning a series of grants is required before we can realize the end goal. An open-source reference implementation that any wallet can use to enable any kind of machine learning on any type of private data, for any purpose they desire. WorkPi will be amongst the first companies to use this in conjunction with the European Blockchain for career related data.

Grant Deliverables:

Grant Deliverable 1:

Create an architecture on how the Compute-To-Data technology could be applied on SSI wallets. Currently Ocean’s compute-to-data uses an architecture that involves kubernetes clusters where a visiting docker container contains the machine learning algorithm and kubernetes configurations allowing it access to (portions) of the data. A new method has to be devised for SSI Wallets as they are typically running on mobile phones which do not allow for apps to run kubernetes server clusters in a usable fashion.

Grant Deliverable 2:

Create a knowledge base with the precise definition of Compute-To-SSI, it’s purpose, architecture (including diagrams) and the way to create software using it. This will be set up as a public git repository and will track further progress of the project as well as allow anyone to participate/contribute.

Grant Deliverable 3:

Model how SSI wallets would interact with the Ocean Protocol ecosystem, i.e. rent existing machine learning models, how to pay for it, or be paid for it. This will naturally be added to the knowledge base.

Grant Deliverable 4:

Model how an interested party can utilize the network of SSI wallets to train their machine learning algorithm for a specific purpose. Define what taxonomies are required. This will naturally also be added to the knowledge base.

Which category best describes your project?

Build / improve applications or integrations to Ocean.

Which Fundamental Metric best describes your project?

Consumer Volume: Compute-To-SSI enables machine learning algorithms to learn from private data that is currently not accessible. This would significantly increase the consumer volume. Also, the open-source reference implementation can be used by other projects to apply machine learning on private data driving adoption further.

What is the final product?

Being able to learn from data residing in SSI wallets that are compliant with European legislation. This means the highest standard of privacy must be implemented.

How does this project drive value to the “fundamental metric” (listed above) and the overall Ocean ecosystem?

There will be at least one new dataset listed which contains the career information of the dataset participants, stored in their SSI wallets. This data will be consumed by a lot of algorithms to generate statistics and machine learning algorithms on top of this data.

To prove and quantify the added value of Compute-To-SSI for the Ocean Ecosystem we have created four different calculations:

  1. Value created & validated by WorkPi with a career data use-case
  2. Total potential value across other use-cases and entities within the eSSIF ecosystem.

1. Value created & validated by WorkPi with a career data use-case

Using WorkPi, people gather career-related data about their skills, characteristics & preferences with assessments or development courses. This type of information can be compared to a LinkedIn profile (although it will be a lot richer because the user is stimulated to track their detailed achievements due to the privacy friendliness).

In 2015 an article revealed that the value of a LinkedIn profile when selling, equals $0,40 while it is being re-sold for ~$200. This means that the intermediary is earning a commission fee of $199,60 which equals an astonishing 49900%. To put this in perspective: currently Apple is under fire for charging a 30% fee in their App Store.

In our calculation we decided to pick an average selling-value of $100, a price that matches the value that we personally received on conferences to share our LinkedIn profile with prominent parties doing data analytics.

Since the buyer does not purchase full access to the profile, but instead only the right to train their A.I. algorithm on the data, we have to take 80% of the profile value because we estimate 60% of the times data is accessed, it is to train an A.I. algorithm. This brings the value of one profile, for machine learning purposes only, to: 0,6 * $100 = $60

WorkPi is currently in the last phase of building the application for our launching client; a large finance company in The Netherlands with ~3.000 employees. Based on this product we are building a SaaS solution for ~40 interested enterprises (which have >1.000 employees in around 14 different industries). The final user group will be onboarded using Work X; our decentralised platform for work where job candidates can participate in the same assessments & development courses as employees. Work X will be launched late 2021 and we expect 2.000-5.000 users in the early phase of the project. Although we expect a large number of users from multiple sources, we are projecting a modest amount of 1.500 users in our calculation.

ROI calculation grants:

* Bang = 1.500 profiles * $ 60 = $90.000
* Buck = $ 17.500
* % chance of success = 80%

Expected ROI = $120.000 / $17.500 * 0,80 = 4,11

2. Total potential value across other use-cases and entities within the eSSIF ecosystem.

WorkPi will participate in an 8-month programme of eSSIF-Lab, supported and funded by the European Commission. Because of large involved governmental stakeholders in this ecosystem (e.g. Spain, Germany, The Netherlands, Belgium), the total potential upside of this grant proposal is very large. The integration with the eSSIF ecosystem is a large process that could take up to 7 rounds.

  • The EU counts around 447 million citizens. In this calculation we project an early adoption rate of 5%.
  • About 86% of these people have access to internet (2018)
  • Although we expect that the value of the data that can be linked to an SSI wallet greatly exceeds that of just the career-related data (given the broad amount of data types), we also use a value of $60 dollars in this calculation.

ROI calculations:

* Bang = 447 million * 0.86 * $ 60 * 0.05 = $ 1.153 billion
* Buck = 7 * 17.500 = $ 122.500
* % chance of success = 70%

Expected ROI = $ / $122.500 * 0,70 = 6590
(yes, that’s a lot… we strongly believe that this could be a billion dollar grant!).

Funding Requested:


Proposal Wallet Address


Have you previously received an OceanDAO Grant?


Team Website:

Twitter Handle:

  • @Rikrapmund
  • @daniel_de_witte

Discord Handle:

  • Rik Rapmund#5093
  • Daniel de Witte#9812

Project lead Contact Email:

Country of Residence:

The Netherlands

Part 2 - Proposal Details

Mockups & Designs

You can find part of our design library here:


  • NodeJS
  • Typescript
  • React
  • Storybook
  • Adobe XD
  • GraphQL
  • Docker
  • Kubernetes
  • Azure
  • Self Sovereign Identity
  • Python
  • Solidity


  • Microservices Framework
  • Continuous Integration
  • Event Based Messaging
  • Flexible Data Model
  • Privacy Oriented Design
  • Recursion
  • Federated Server System
  • Isomorphic Role-based Authorization

Project Deliverables - Roadmap

Beyond the scope of this grant proposal we have designed a roadmap to show the long term goal of the entire project:

  • Identify open source SSI wallets that are compatible with EBSI and have sufficient adoption to succeed.

  • Align with the wallet dev-teams how to implement receiving ocean algo’s and training them.

  • Training with Zero Knowledge (For example Apple’s FaceID faced acceptance problems, even though only the weights of the trained model were sent back to the server)

  • Create a security model that can guarantee the privacy of an individual’s data while being visited by machine learning algorithms. (Zero Knowledge Proofs could help here)

  • Create a reference implementation in one of the chosen wallets.

  • Keep expanding wallet support to grow the ecosystem of wallets able to host Ocean machine learning algorithms.

  • Utilize A.I. chips in phones to speed up calculations, reduce energy consumption and keep your phone cool.

Team members

  1. Rik Rapmund
  1. Daniel de Witte
  1. Darius Costolas - Full-stack developer

  2. Dragos Podaru - Front-end developer

  3. Pavlik Kiselev - Quality Assurance

  4. Alex Basmanov - Front-End developer

  5. Tiberiu Paliuc - Graphic designer

  6. Hidde Kehrer - Operations & Funding

  7. Kirill Ryadchin - Junior front-end developer

Additional Information

  • WorkPi participates in the eSSIF development trajectory from the European Commision to enable individuals to regain control over their work related data using Self Sovereign Identity.

  • WorkPi has an enterprise launching client called APG.

  • WorkPi is a member of the Dutch Blockchain Coalition.

  • WorkPi is an Odyssey 2020 hackathon winner in the SSI category.

  • WorkPi has been featured on Dutch national television explaining the project in detail:


Pleased to meet you all!
This is Daniel of WorkPi / Work X. I’d be happy to answer any questions.


Welcome onboard and nice chatting to you earlier!


Hello from ResilientML

This looks like a strong and well thought out project - welcome to Ocean community from ResilientML. I look forward to seeing this interesting solution evolve and the use-case proposed for job markets as a starting point looks interesting.

It is reminiscent to me of projects such as SelfKey - how do you compare to such tech being created for similar use cases?

I would like to ask a few questions on your deliverables:

Deliverable 1: Do you have a clear technology prototype already in consideration for this solution - or is this more bluesky in nature - it seems like this is a reasonable tech challenge.

Deliverable 2 and 3: - sounds great - look forward to reading this as it evolves.

Deliverable 4: specifically interested here in your thoughts on how you will split ML solutions or models to taxonomies - will it be based on the nature of the ML solution architecture, how distribute they are, whether they run minibatching or fine-tuning or the cost of computing?

Overall - a very interesting proposal - look forward to your thoughts

thanks ResilientML.


I like where this is heading and will most likely support this. Thank you for joining the community.


Thank you all for the warm welcome!

To answer your questions Gareth we would have to look closely at the goal we are trying to achieve: building a reference implementation on an existing opensource SSI wallet that is modular and can be reused by anyone.

With that in mind lets dive into the deliverables:

Deliverable 1: At WorkPi we want to use SSI and still be able to learn from it, however we do not want to reinvent the wheel nor build yet another SSI wallet. SelfKey is offering their own SSI Wallet and their technology is focussed/limited to their own app.
We want to leverage existing SSI Wallets that are open source, and build a reference implementation that allows an A.I. to learn from the data within, without violating privacy. We will be amongst its first users, however since the implementation is generic/modularized, other SSI wallets would be free to use/copy it, and we would encourage them to do so.
I think this is a significant difference from what SelfKey, and many other SSI wallets are trying to do, in fact if their software would be open-source, which it seems to be, they would be a potential candidate for our reference implementation.

Deliverable 2 and 3: Well thank you, I am just as much looking forward to it and hope it will get good traction if other projects start to realize the potential it could have for them.

Deliverable 4: Here the same philosophy applies, we are not trying to re-invent the wheel, but use the best of what is already there and improve/combine it in order to reach our goal. As taxonomy we plan to use an European standard called ‘Europass’. They have been laying out the groundwork, for what we think will be the golden standard for identity, qualification and career related data. We want to augment what is already there and make it better.

About the ML Solution architecture this is part of the process of finding out what is best by collecting knowledge from many qualified sources and combining this, until these experts reach some form of consensus. That being said, we do believe you are right that mini-batching will enable a very important feature, namely that the person in control of the wallet is able to give and/or revoke access to specific elements of their data. Regarding fine-tuning, we think this could greatly optimize performance, however we need to see if/how it works in conjunction with federated learning as well as differential privacy which is required to avoid fingerprinting privacy concerns.

I hope I have answered your questions to your satisfaction and thanks again for the warm welcome.

1 Like

Dear WorkPi

Great directions and your response makes a lot of sense - I look forward to hearing more specifics of the details related to my questions arise as you progress and hopefully ResilientML will then be able to find ways to engage with your solution and the work we are doing on FaaS and ML solutions for Compute-to-data can then eventually take advantage of innovations and implementations you may bring to Ocean - great to meet you and thank you for taking the time to respond carefully.

best wishes