This paper presents a method for forecasting essential project metrics such as cost, schedule, and risk factors using long short-term memory (LSTM). We demonstrate that LSTMs effectively capture the relationships between various aspects of project management. We generate a diverse synthetic dataset that simulates different project scenarios and subsequently develop an LSTM model to learn multiple metrics simultaneously, including project duration, cost, and the probability of overruns or disruptions. We conducted experiments to evaluate whether our LSTM-based approach provides reliable predictions, offering project managers a data-driven means to plan resources, schedules, and risk responses. We also propose suggestions for future work, including validation with actual project data and the integration of additional project factors.