ZK ≠ Zero Income

This project aims at offering a proof of concept for a pluggable proof of income utilizing zero knowledge and AI for privacy.. The user inputs an image of its income sheet and the neural network is able to determine in which category of income the user belong. Then the user can submit a proof to the verifier on chain to prove that they have a salary large enough for, for instance, applying for housing, without revealing the exact amount.

This project consists of a neural network made with python and pytorch, able to classify incomes from an income sheet image into three different categories. The input and output of the inference of the network is then used to build a verifier contract, deployed on-chain and to compile proofs of income in which only the output of the inference is public. The verifier contract is then used in a very simple contract to gate the switch of a isVerified boolean which can be only switch to true if the user submits a valid proof.