• Pythia: A machine learning experiment management platform with a companion Python module. Gives overview of experiments containing models, which have metrics, hyperparameters and so on. Aims to keep track of the things you tried and helps you identify what worked.

  • Kvasir: A HTTP micro microservice registry written in Node.js with Lokijs as in memory database.

  • Numerflow: A luigi data workflow collection for the machine learning competitition. Supports automatic data set updates, model training and submission with a complete pypeline.

  • Sparkle.js: Sparkle update framework for OS X backend aggregrating anonymous feedback about the computers as well as deploying future updates.

  • A DIY cloud for everyone to host with native apps that allow you to just drop some file on the app icon and get a URL for the file pasted to your clipboard. Sharing made easy, without disclosing your files. We have a C89 API library, an OS X app, a Node.js backend and hope to fill the gap for some other devices in the future too. Drop me a line if you want to help out on this. The project on GitHub.

  • libdeluge-c: (Halted) I started to work on this project only recently and it is not really advanced. Deluge is a torrent daemon with an RPC API which I want to make consumable with this lib.

  • HappyTypo: Angular module fixing micro typography issues for German text. Pretty much left alone quite some time.