Master AI Agents Development in just 4 weekends

Theory (1 hour):

• • Quick refresher on Transformer architecture
• • Pre-training and fine-tuning of large language models (LLMs)
• • Key differences between generative AI and agentic AI

Hands-On Lab (1 hour):

• • Setting up Python environment (Anaconda, venv)
• • Experimenting with OpenAI API and local models
• • Writing scripts to query LLMs and parse responses

Theory (1 hour):

• • Types of planning: reactive vs. deliberative in AI agents
• • Structured chain-of-thought prompting techniques
• • Multi-step reasoning and error correction strategies

Hands-On Lab (1 hour):

• • Implementing chain-of-thought prompting
• • Designing reasoning workflows for logic puzzles
• • State tracking between LLM queries

Theory (1 hour):

• • Short-term vs. long-term memory in AI agents
• • Vector databases and knowledge graphs
• • Retrieval-Augmented Generation (RAG) architecture

Hands-On Lab (1 hour):

• • Setting up FAISS/Pinecone vector stores
• • Building RAG-powered agents
• • Evaluating retrieval quality metrics

Theory (1 hour):

• • Agent-to-tool interaction patterns
• • Reinforcement Learning approaches for agents
• • Policy gradients vs. Q-learning overview

Hands-On Lab (1 hour):

• • Connecting agents to external APIs
• • Basic RL environment implementation
• • LLM-based tool selection logic

Theory (1 hour):

• • Advanced prompt construction techniques
• • Automatic prompt optimization
• • Self-debugging and reflection strategies

Hands-On Lab (1 hour):

• • Multi-layer prompt design for complex tasks
• • Building self-improving feedback loops
• • Logging and analytics implementation

Theory (1 hour):

• • Web parsing fundamentals: DOM, HTML, CSS
• • Selenium vs. Playwright comparison
• • Session management and dynamic content

Hands-On Lab (1 hour):

• • Setting up Playwright/Selenium
• • Automating form filling and data extraction
• • AI-driven browser interactions

Theory (1 hour):

• • Multi-agent system architectures
• • Communication protocols between agents
• • Error handling and fallback strategies

Hands-On Lab (1 hour):

• • Setting up specialized agent teams
• • Orchestration layer design
• • Robust error detection implementation

Project Overview:

Build a fully autonomous FinTech AI agent that analyzes financial data and generates market research reports.

Key Components:

• • Web scraping from Yahoo Finance & Google News
• • Multi-agent system using Autogen framework
• • PyCharm development environment setup
• • Streamlit frontend implementation
• • Google Cloud deployment