At ScAI, we are excited about developing AI-based solutions to sequential decision making problems under uncertainty. This is especially important given the increasing prevalence of AI agents in both simulated and physical contexts. We apply techniques from reinforcement learning. Some recent directions include using RL in problems related to multi-agent systems, fine-tuning of foundation models, deciding optimal actions for embodied agents in defense, autonomous driving and industry, and RL combined with safe AI. We also work on developing new RL algorithms as well as finding novel ways to combine RL with ideas from closely aligned research thrusts on deep learning, generative models, and explainability