We are currently developing a set of environments to test various AI algorithms. The AI algorithms work with various environment representations, so every environment has several various representations for different AI algorithms. The thesis topic is to create an executor service to run experiments with such environments. A typical experiment specification consists of a list of AI algorithms and environments to try those algorithms with. The executor then runs each algorithm in each environment several times, collects logs, experiment results and statistics. Multiple experiments should be allowed to run at the same time, if there are enough hardware resources. The executor has a web interface for configuring experiments and scheduling them for execution and for retrieving experimental results.