Module 1: Introduction to Agentic AI

• 1.1 What is Agentic AI?

  • Definition and key characteristics

  • Differences between reactive, deliberative, and agentic AI

• 1.2 Historical Context and Evolution

  • Early AI agents and expert systems

  • Rise of autonomous systems and multi-agent environments

• 1.3 Applications of Agentic AI

  • Robotics, autonomous vehicles, virtual assistants

  • AI in strategic decision-making and simulations

Hands-on: Explore simple agent simulations using OpenAI Gym

Module 2: Core Principles of Agentic AI

• 2.1 Agent Architectures

  • Reactive, model-based, hybrid, and goal-driven architectures

• 2.2 Decision-Making and Planning

  • Markov Decision Processes (MDPs) and Partially Observable MDPs (POMDPs)

  • Planning algorithms: A*, Monte Carlo Tree Search, and hierarchical planning

• 2.3 Learning in Agents

  • Reinforcement learning, imitation learning, and offline learning

Hands-on: Implement a basic RL agent to solve a navigation problem

Module 3: Advanced Agentic AI Techniques

• 3.1 Multi-Agent Systems

  • Cooperation, competition, and communication strategies

• 3.2 Human-AI Collaboration

  • Assistive agents, recommendation agents, and human-in-the-loop systems

• 3.3 Emergent Behaviors

  • Exploration of self-organization and adaptation in agentic systems

Hands-on: Create a multi-agent environment with cooperative and competitive agents

Module 4: Safety, Alignment, and Ethics

• 4.1 AI Safety Challenges

  • Goal misalignment, reward hacking, and unintended behaviors

• 4.2 Ethical Considerations

  • Bias, fairness, and accountability in autonomous systems

• 4.3 Governance and Policy

  • Regulatory frameworks for autonomous AI systems

Hands-on: Evaluate an agentic AI system for ethical risks and propose mitigations

Module 5: Tools and Frameworks

• 5.1 Development Frameworks

  • OpenAI Gym, RLlib, PettingZoo, Unity ML-Agents

• 5.2 Simulation Platforms

  • Virtual environments for training and testing agents

• 5.3 Monitoring and Debugging Agents

  • Metrics, logging, and interpretability

Hands-on: Build, train, and monitor a simulated agent in a complex environment

Module 6: Real-World Projects

• 6.1 Autonomous Navigation Agent

• 6.2 Resource Allocation and Optimization Agent

• 6.3 Human-AI Collaboration Agent

• 6.4 Ethical and Safe Agent Deployment Case Study

Course Delivery Mode

• Lectures: Concepts, architectures, decision-making, ethics

• Labs: RL agents, multi-agent systems, simulation environments

• Assignments: Small exercises reinforcing theory and coding

• Capstone: Full agentic AI project integrating learning

Prerequisites for joining the course:

• Basic understanding of AI/ML concepts

• Programming proficiency (Python recommended)

• Familiarity with reinforcement learning and decision-making models

Capstone Project:

• This course balances theory, practice, and ethics, with two major hands-on projects: one early reinforcement learning exercise and a capstone project at the end.
• Capstone Project-Design and implement a fully agentic system addressing a real-world problem, with attention to performance, safety, and ethical considerations

Outcomes of the course :

• Basic understanding of AI/ML concepts

• Programming proficiency (Python recommended)

• Familiarity with reinforcement learning and decision-making models

Industry specific Training

The program can be tailored for corporate training, allowing organizations to implement industry-specific agentic AI solutions. Participants will gain hands-on experience in building autonomous AI agents that address real business challenges, improve decision-making, and drive innovation within their specific domain.