This repository provides hands-on implementations of the four agentic patterns defined by Andrew Ng’s DeepLearning.AI curriculum. It covers the reflection pattern where LLMs self-review their outputs, the tool-use pattern for enabling agents to call external functions, the planning pattern using ReAct-style reasoning, and the multi-agent pattern for coordinating multiple agents in a crew. All implementations are built from scratch using direct Groq API calls without relying on high-level frameworks like LangChain or LangGraph.