Current Projects

XCLAIM: Research on Blockchain Interoperability

XCLAIM is a protocol that enables cross-chain interoperability of cryptocurrencies. The idea is to issue tokens from one blockchain on another chain without any trusted intermediaries. Instead, the intermediaries are incentivized to behave honestly through collateral. Proofs of honest and malicious behaviour is achieved through transaction inclusion proofs. This is the first trustless cross-chain interoperability protocol with such strong guarantees and without the need to either trust a third-party or bootstrap a permissioned blockchain as intermediary.

A specification language for smart contract verification

Smart contracts are still prone to security vulnerabilities. Apart from using automated tools to verify for known bugs, it can be beneficial to verify that a smart contract is correct with respect to some specification. However, using formal verification tools is mostly combined with a steep learning curve and requires a deep understanding of the tools. This projects aims to develop a domain-specific language for writing specifications to verify smart contracts. The idea is to abstract away some of the complexities so that formal verification can become a standard tool for smart contract developers.

Past Projects

Privacy-preserving machine learning

I had the pleasure to supervise Florian Apfelbeck's master thesis project to develop a privacy-preserving machine learning project. In this project, a Multi-Party Computation (MPC) protocol was used to realise a linear regression and a shallow neural network implementation without reveling input data of the participants in the protocol.

dInvest: a social hedge fund in Ethereum

Selecting a suitable financial investment, ethical and social criteria are becoming critical success factors. The core issue is finding a suitable framework for sustainable and profitable investment in combination with full transparency over the decision process and transactions. We are working on a proof-of-concept to realize profitable, transparent and socially responsible investment.

Breaking CAPTCHAs using deep learning

The aim of the project is to break CAPTCHAs using deep learning technologies. Initially the focus lies on simple CAPTCHAs to evaluate the performance and move into more complex CAPTCHAs. The training dataset is generated from an open source CAPTCHA generation software. Tensorflow is used to create, train and test the network.