With launch services getting affordable the big players in the satellite and commercial space industries are looking for ways to deploy and maintain large satellite constellations. It is projected that in the next decade the number of satellites in the earth's orbit will increase from a few thousand to a hundred thousand. Manually managing such a large number of satellites with conventional control centers and workers will be very hard and expensive hence operators and space agencies are looking for ways to automate the constellation management efforts. Artificial intelligence (AI)  and machine learning algorithms can support those efforts. Collision and debris avoidance, the establishment and maintenance of the communication network, health of the satellite, and its operational routine are a few of the tasks which we are targeting to solve with the use of AI. Our solution will use data from hundreds of constellations forming satellites already in orbit as an input for machine learning algorithms. The resulting model will be used to build the scale-up simulation with hundreds of thousands of satellites and use the AI to achieve the autonomy of the constellation.