entbappy/complete-agentic-ai-course

Repository Scope

This is a course-material monorepo for agentic-AI concepts rather than one deployable application. It combines notebooks, PDFs, and several standalone Python examples covering LangChain agents, LangGraph workflows, validation, asynchronous programming, and interview preparation.

Primary Entry Points

Runnable project entry points are concentrated in:

• LangChain-Multi-Agent-Research-System/app.py and LangChain-Multi-Agent-Research-System/main.py

• Langchain-Single-Agent/app.py and Langchain-Single-Agent/main.py

• LangGraph-Code/Agentic-Chatbot-using-LangGraph/app_simple.py

• Variant chatbot UIs: LangGraph-Code/Agentic-Chatbot-using-LangGraph/app_db.py, app_rag.py, app_hitl.py, and app_tool.py

The numerous LangGraph-Code/*.ipynb files are instructional, independently executable workflow examples rather than modules in a single runtime.

Multi-Agent Research System

LangChain-Multi-Agent-Research-System is the most conventionally layered application:

• Agent definitions and delegation logic: src/agents/agents.py

• Research/tool integrations: src/tools/tools.py

• Pipeline orchestration: src/pipelines/pipeline.py

Control flow is expected to enter through the top-level app scripts, construct agents/tools, and execute the pipeline. The src boundary separates orchestration, agent behavior, and external-capability adapters.

Single-Agent LangChain Example

Langchain-Single-Agent/main.py and Langchain-Single-Agent/app.py provide a smaller standalone agent application. Supporting experimentation is isolated under Langchain-Single-Agent/research/agent_demo.ipynb. This forms the simpler baseline against which the multi-agent system’s delegation and pipeline layering can be understood.

LangGraph Workflow Curriculum

LangGraph-Code contains progressive examples of graph-based stateful control flow: prompt chaining, moderation, review, iteration, persistence, human-in-the-loop (HITL), subgraphs, shared subgraphs, and guardrails. Representative stages are in:

• LangGraph-Code/3_Prompt_Chaining_Workflow.ipynb

• LangGraph-Code/8_Iterative_Workflows.ipynb

• LangGraph-Code/9_Persistence.ipynb

• LangGraph-Code/10_HITL.ipynb

• LangGraph-Code/11_subgraphs.ipynb

• LangGraph-Code/13_guardrails_crash_course.ipynb

These are educational graph patterns, not a shared package dependency tree.

Agentic Chatbot Variants

LangGraph-Code/Agentic-Chatbot-using-LangGraph is a feature-oriented chatbot implementation with alternate graph backends:

• Basic conversation: agentic_chatbot_backend.py

• Tool-enabled agent: agentic_chatbot_tool_backend.py

• RAG-enabled agent: agentic_chatbot_rag_backend.py

• Persistent database-backed agent: agentic_chatbot_db_backend.py

• Human-approval/interrupt path: agentic_chatbot_hitl_backend.py

Each corresponding app_*.py file is a presentation/runtime entry point. The boundary is clear: app scripts collect/display conversation interactions; backend scripts define LangGraph execution behavior.

State, Retrieval, and External Dependencies

The chatbot persists local conversation-related state in SQLite artifacts: chatbot.db, chatbot.db-wal, and chatbot.db-shm. Its RAG path uses a local FAISS vector index at faiss_db/index.faiss and faiss_db/index.pkl. Python dependencies are scoped per subproject in LangGraph-Code/requirements.txt, LangGraph-Code/Agentic-Chatbot-using-LangGraph/requirements.txt, and the LangChain project requirements files. Environment files in LangGraph areas indicate external LLM/API credentials, but no cloud deployment infrastructure is represented in the tree.