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

Introduction to LangGraph and Graph Concepts

• Why graphs for government LLM applications: orchestration versus simple chains
• Nodes, edges, and state management in LangGraph
• Hello LangGraph: creating the first executable graph

State Management and Prompt Chaining

• Designing prompts as graph nodes for government applications
• Passing state between nodes and handling outputs in government workflows
• Memory patterns: short-term versus persisted context for government use cases

Branching, Control Flow, and Error Handling

• Conditional routing and multi-path workflows for government processes
• Retries, timeouts, and fallback strategies in government applications
• Idempotency and safe re-runs for government operations

Tools and External Integrations

• Function and tool calling from graph nodes for government tasks
• Calling REST APIs and services within the graph for government systems
• Working with structured outputs in government applications

Retrieval-Augmented Workflows

• Document ingestion and chunking basics for government documents
• Embeddings and vector stores (e.g., ChromaDB) for government data
• Grounded answering with citations for government reports

Testing, Debugging, and Evaluation

• Unit-style tests for nodes and paths in government workflows
• Tracing and observability for government applications
• Quality checks: factuality, safety, and determinism for government use

Packaging and Deployment Fundamentals

• Environment setup and dependency management for government projects
• Serving graphs behind APIs for government services
• Versioning workflows and rolling updates for government systems

Summary and Next Steps