The problem Sentinel.ai, Incentivised ML on Blockchain. solves

1. Availability of good organic data sources is one of the primary problems with AI. With advances in scalablity, blockchain can be utilized as backbone to facilitate decentralized machine learning technologies.
2. Sentinel.ai allows these data contributers to monetize their data by providing it towards training a ML Model.
3. Sentinel.ai allows any ML model to be trained accross the network for a cost proportional to the number of rounds and complexity of the model. This allows us to reward users for their contribution.
4. The system is built on the federated learning architecture with local differential privacy, allowing us to utilize user data with revealing any information about them.
5. Sentinel.ai makes it safe for users to contribute their data, as their personal data never leaves the device and all the inferences made from their data are encrytped, leaving no way for the information to trace back to the user.
6. ML models can be trained in a decentralized manner by the anyone, coordinated using a smart contract ensuring users are rewarded for their work.
7. Sentinel.ai improves the accuracy of the trained model by brining in richer organic data and at the same time preserves user privacy and with Incentivised Machine Learning on Blockchain allowing us to reward users based on the quality and quantity of their contribution.
8. Sentinel.ai prevents adversaries from messing with the training by measuring the privacy loss and the accuracy of the model before broadcasting the model gloablly.
9. Sentinel.ai opereates in a completely decentralized manner where anyone can join the network and get rewarded in a safe and robust manner.
10. With an intuitive interface phasing out the complexity behind the network, Sentinel makes it incredibly easy for anyone to get started training with a simple drag and drop interface.

Sentinel.ai, Democratized and Incentivized Machine Learning on Blockchain in a Secure and Privacy-Preserving Manner.

Challenges I ran into

Handling coordination of training a model across a decentralized set of nodes was a challenge with I was able to solve by bringing even the coordination over blockchain, increasing transparency and improving the security of the network.