The increasing adoption of cloud-edge-IoT infrastructures has transformed modern distributed systems into highly heterogeneous and dynamic ecosystems spanning the entire computing continuum. Applications must operate across geographically distributed resources, heterogeneous hardware platforms, and rapidly changing workload conditions. In these environments, resource management decisions including scheduling, orchestration, placement, scaling, migration, and load balancing must continuously adapt to fluctuating demands while satisfying performance, cost, energy, and quality-of-service objectives. Traditional optimization and rule-based approaches often struggle to cope with the complexity, uncertainty, and scale of these systems. Reinforcement Learning (RL) has emerged as a powerful paradigm for designing self-adaptive resource management mechanisms capable of learning effective decision-making policies directly from interaction with the environment. More recently, Multi-Agent Reinforcement Learning (MARL) has attracted significant attention as a means to address decentralized control problems involving multiple interacting components operating across cloud, edge, and IoT layers. This tutorial provides a comprehensive introduction to RL and MARL techniques for intelligent resource management in distributed systems and the computing continuum. The tutorial combines theoretical foundations with practical experience, enabling participants to understand the principles behind learning-based control and to apply state-of-the-art tools to real resource-management scenarios.

Prerequisites: Basic knowledge of distributed systems and cloud/edge computing concepts; familiarity with Python programming; basic understanding of machine learning concepts is beneficial but not required. No prior experience with Reinforcement Learning is assumed.