The Protocols and Structures for Inference project aims to develop a general purpose web API for machine learning. This small web.py service provides a mechanism for training and obtaining predictions with scikit-learn as well as an implementation of a simple ranking algorithm based on logistic regression. It is not a PSI service, in that its API does not follow the PSI specification, but it can be used by an implementation of that specification to interact with scikit-learn.
Further documentation may be provided in the future.