Companies are constantly hiring new employees. To efficiently allocate these newcomers to tasks, we need to predict which tasks they can do and how well they will perform on these tasks. However, making such predictions for newcomers is challenging, particularly at the early stage of onboarding, due to the limited availability of observational data on their past experience. In this research, we explore the problem of predicting newcomer capabilities and performance in the context of process execution and propose a solution to address the challenge. The proposed approach uses data augmentation from historical event data, guided by organizational model mining, and generates predictions for the tasks that newcomers may perform and the time it will take them to perform those tasks. Experiments based on several real-life event logs showed that the proposed approach achieves accurate predictions for newcomers given limited data availability.