Hi! I'm a Full Stack DeveloperClick me!

Building WCAG accessible software for the web

For 8 years I helped deliver the WISEflow digital assessment platform. I was responsible for developing the frontend together with a team of developers. During the development, I became one of the developers with the most knowledge about the platform and its many facets. The code was built with React/Typescript and Golang, but developed on top of a large legacy codebase constisting of PHP, AngularJS, JQuery and more.

As a Computer Science master's student I focused on Cryptography. Given my technical skills and interests it was a great opportunity to work on a project that involved the implementation of a post-quantum digital signature protocol in Rust. The project gave me the opportunity to learn about the implementation of underlying cryptographic protocols like Secret Sharing, MPC and Code-based cryptography.

As a senior developer in the organisation I took a major part in developing and maintaining the WISEflow marking tool from the frontend side. In collaboration with the team we develpoed a micro frontend SPA in React and took use of modern tools like Tanstack/React-Query and PDFtron.

As a developer and Frontend Tech lead I worked both as a contributor and administrator of the UI component library used by the WISEflow React frontend. I was responsible for developing well designed, accessible and reusable components to be used in the development of modern parts of WISEflow. I took a leadership position in driving the direction of the library and handling the release structure

Abstract - Scientific research is currently under much pressure from the scientific community as recent studies have indicated that common methods do not give replicable findings. The criticism has recently accelerated with the “replication crisis” of Psychology and the increasing frustration with publication bias. With its increase in popularity, meta-analysis has recently stood as a plausible solution to this crisis of scientific method. However, there are still issues that meta-analysis seem to struggle to solve. With the increase in computational power, Bayesian ideas pose radically different approaches that could solve many of the crisis issues elegantly, but Bayesian methods are still underutilised in meta-analysis. This paper introduces and overviews the BHLM package to R, which utilises hierarchical latent mixture analysis as an approach to meta-analysis. Using a prototype build of the package, I re-analyse two recent meta-analysis studies. The analyses produce negligible effects, similar to the original papers, which supports the validity of the package and the method. However, the method allows for more elaborate analysis based on further developed and theorised priors. In the end, future implementations to the package are discussed. These include, Bayesian bias mitigation and assessment and the package as a stepping tool to introduce Bayesian methods to more researchers. Such introductions, ideally, could lead towards a solution to the current scientific crisis.

As a project for the Cryptographic Computing Course at Aarhus University, I collaborated on a project to implement a Threshold Signature Scheme using Multi-Party Computation.

Abstract - With recent developments in computer science, we propose that elections can benefit from digital solutions that utilise cryptographic methods. With inspiration from auction applications developed with Multiparty Computation (MPC) by Bogetoft et al., we set out to implement a digital election system that upholds the important characteristics of an election, i.e. privacy and integrity, with MPC in an active security setting. We describe current theories of MPC, Secret Sharing, and Distributed Systems, along with schemes that provide integrity to the system in the form of verifiability and agreement. We provide two implementations: an implementation with passive security to provide a baseline for comparison and an active implementation. Benchmarks provided valuable insight into the performance of our election system and small changes gave dramatic performance increases. While the system still requires a considerable of research, it provides a preliminary solution to elections in terms of performance and security.