Traditional automation technologies alone are not sufficient to enable driverless operation of trains (called Grade of Automation (GoA) 4) on non-restricted infrastructure. The required perception tasks are nowadays realized using Machine Learning (ML) and thus need to be developed and deployed reliably and efficiently. The safe.trAIn project (2022 - 2024) aims to lay the foundation for safe use of ML for the driverless operation of rail vehicles and to thus addresses key technological challenges hindering the adoption of unmanned rail transport. Therefore, the project investigates methods to assess trustworthiness of ML models taking robustness, performance, uncertainty, and transparency aspects of the ML model into account. These methods must be integrated into an MLOps process for tackling improved reproducibility, traceability, collaboration, and continuous adaptation of the autonomous operation to changing conditions. MLOps mixes ML application development and operation (Ops) and enables high frequency software releases and continuous innovation based on the feedback from operations. In this talk, we present challenges and outline solution strategies to setup an MLOps process for the continuous development and safety assurance of the ML models to realize the obstacle detection functionality of a driverless regional train. This process integrates system & software engineering, safety assurance, and ML engineering in a comprehensive workflow. We present the individual stages of this process and how the different activities interact. Moreover, we describe relevant challenges and solution strategies to automate the different stages of the MLOps process.