We like building and contributing to open source software, and believe in the open source movement. The following are open source projects we’ve created and maintain. If you’d like to contribute or want to chat about any of these projects, drop us an email.
Phoenix
An open-source social media analysis platform for peacebuilders.
Phoenix is a powerful, multi-tier platform that enables peacebuilders and analysts to monitor, analyze, and visualize social media data for conflict prevention and peacebuilding efforts. Built in collaboration with Build Up.
Since January 2025, Build Up offers a managed platform that is free to use for peacebuilders. Learn more at Build Up Phoenix.
License: GNU AGPLv3
Kotsu
kotsu is a Python package that provides a lightweight and flexible framework to structure validating and comparing machine learning models.
It aims to provide the skeleton on which to develop models and to validate them in a robust and repeatable way, minimizing bloat or overhead. Its flexibility allows usage with any model interface and any validation technique, no matter how complex. The structure it provides avoids common pitfalls that occur when attempting to make fair comparisons between models.
License: MIT
Truman
Dynamic complex-system simulations for one-shot optimal decision-making agents.
truman is a Python package that implements suites of environments (system simulations) exhibiting behaviours of real world large scale systems, e.g. changes in online consumer cohort conversion as a function of changes in product price.
truman is not an environment for training reinforcement learning agents, but aims to be an effective way to develop and validate one-shot optimal decision-making agents that perform well on unique systems that can’t be reliably simulated and that have a high cost of experimentation.
License: MIT
Pachinko
One-shot optimal decision-making agents and algorithms.
pachinko is a package that implements one-shot optimal decision-making agents to solve environments provided by the truman package and compatible with OpenAI Gym environments. The goal is to implement agents that learn in a single experience, so that they can perform well on unique systems that can’t be reliably simulated and that have a high cost of experimentation.
License: MIT