notes-ai-agents

Overview

This repository contains a series of notebooks that progressively build and evaluate LLM-powered chatbot, rag and agents systems. The focus is on experimentation with safeguards, observability, and evaluation frameworks.

In /checklist folder:

checklists
├── 0_tool_design.ipynb
├── 1_metrics_eval_agents.ipynb
├── 2_safety_measures.ipynb
└── 3_deployment.ipynb

In /steps folder:

steps
├── text
├── vision
│   └── imgs
└── voice

• /text

  text
  ├── 0_basic_agent.ipynb
  ├── 1_agent_confidence.ipynb
  ├── 2_ollama_websearch.ipynb
  ├── 3_small_chat_on_sustainability.ipynb
  ├── 4_ollama_safeguard.ipynb
  ├── 5_llm_mlflow_tracing.ipynb
  ├── 6_chabot_mlflow_tracing.ipynb
  ├── 7_chatbot_evaluation.ipynb
  ├── 8_compare_llm.ipynb
  ├── 9_streamlit_fastapi_mlflow.ipynb
  ├── 10_token_budget.ipynb
  ├── 11_semantic_cache.ipynb
  ├── 12_semantic_route.ipynb
  └── 13_prompt_builder.ipynb

• /vision

  vision
  ├── 0_vlm_call.ipynb
  └── imgs
      └── car.jpeg

In /advanced folder:

advanced
├── agent
└── rag

• /agent:

  agent
  ├── 0_react_agent.ipynb
  ├── 1_memory_agent.ipynb
  ├── 2_reflection_agent.ipynb
  ├── agent_trace_logs.jsonl
  └── utils
      ├── logger.py
      └── token_budget.py

• /rag:

  rag
  ├── 0_basic-rag-n8n
  │   ├── README.md
  │   ├── articles
  │   ├── docker-compose.yaml
  │   ├── img
  │   │   ├── rag-workflow.png
  │   │   └── text-classifier.png
  │   └── template
  │       ├── simple-RAG-workflow.json
  │       └── text-classifier-ingestion.json
  ├── 1_kg_rag
  │   ├── 0_knowledge_graph_rag.ipynb
  │   └── README.md
  └── imgs

License

This project is under the MIT license.