Glint is a high performance Scala parameter server, compatible with Spark. It was developed as part of my Master's thesis at ETH Zürich.
GlintLDA is a distributed implementation of LDA that uses a Metropolis-Hastings based collapsed Gibbs sampler to speed up computation. This implementation was used to compute a topic model on the full ClueWeb12 data set (20+ TB) in under 80 hours on a cluster of 30 machines.
Shoelace is a neural list-wise learning-to-rank library for Chainer. The library contains functionality that is useful for running learning-to-rank experiments. Implementations of ListNet and ListMLE are included.
Murmur3.jl is a pure Julia implementation of the very fast Murmur3 hashing algorithm. The performance of this implementation is close to that of the default C++ version of Murmur3.