The modern agile software development process relies on task tracking systems, responsible for organizing the development process and allocating the workload to the team members. Assigning tasks to the most suitable member (triaging) is a critical and demanding process and is implemented by evaluating the features of a task report (title, description, labels, importance etc.). Previous attempts to tackle the complicated and time-consuming problem of triaging are usually limited to bug reports analysis. This dissertation introduces a method of automating the triaging process, without constraints on the type of task. Specifically, it aims to investigate the possibility of predicting the most suitable programmer to complete a task just by using the task report. Data from different repositories are used, with the main focus on text data (title, description, labels). Our method is based on applying text processing and data analysis techniques. In contrast to existing research, limited to simple text preprocessing methods (tokenization, lemmatization etc.), topic mod elling techniques (LDA) are also applied in order to extract the topics of each report and enhance its labels. Finally, data are broken down into training and test sets and are used as input for classification models (Naive Bayes and SVM). The pro posed method proved effective, accurately assigning tasks, with the topic modeling techniques contributing significantly to efficiency improvement.