LangGraph Foundations: Graph-Based LLM Prompting and Chaining Training Course

Introduction to LangGraph and Graph Theory

• The rationale for using graphs in LLM applications: orchestration vs. simple chaining
• Understanding nodes, edges, and state within LangGraph
• Getting started: Creating your first executable graph in LangGraph

State Management and Prompt Chaining

• Structuring prompts as individual graph nodes
• Managing state transfer between nodes and processing outputs
• Memory strategies: Differentiating short-term context from persisted data

Branching, Control Flow, and Error Management

• Implementing conditional routing and multi-path workflows
• Handling retries, timeouts, and fallback mechanisms
• Ensuring idempotency and safe execution of re-runs

Tools and External Integrations

• Executing function and tool calls from graph nodes
• Invoking REST APIs and external services within the graph structure
• Managing structured output formats

Retrieval-Augmented Workflows

• Basics of document ingestion and chunking
• Working with embeddings and vector stores (e.g., ChromaDB)
• Generating grounded answers with accurate citations

Testing, Debugging, and Evaluation

• Writing unit-style tests for individual nodes and entire paths
• Utilizing tracing and observability tools
• Performing quality checks on factuality, safety, and determinism

Deployment and Packaging Essentials

• Setting up the environment and managing dependencies
• Serving graph applications via API endpoints
• Implementing workflow versioning and rolling updates

Conclusion and Future Directions