Dengulata

LLM/SLM helper codes and notes

ToDos

• Try Huggingface
• Try MCP with Ollama either OpenAI sdk or LangChain
• Try Claude MCP
• Try RAG with Langchain
• Try LangChain and LangGraph
• Create notes on running docker chatgpt like ui for chatting with all your APIs
• LangChain chain of thoughts using Gemini
• ElevenLabs, Whisper(Speach to text), xtts2 or Openvoice(Text-to-Speech)
• Streamlet in py and gradio
• course
  • https://app.pluralsight.com/library/courses/introduction-developing-ai-agents/table-of-contents, https://github.com/huggingface/agents-course
  • GenAI Essentials – Full Course for Beginners
  • LLM Course – Build a Semantic Book Recommender (Python, OpenAI, LangChain, Gradio) | code
  • AI Engineer Roadmap – How to Learn AI in 2025

Services to work with LLM models

1. Ollama
2. Huggingface
3. AI Cloud serivces - Microsoft AI Foundary (Azure), AWS Bedrock
   • AI Agents in Azure

Interesting models to try

1. Phi3.5 - Tiny model that can be run on CPU
2. LLama3.2 - LLama model can be run on CPU
3. gpt-4o-mini - Cheapest API
4. gemini-2.0-flash - Offers free API

Important practical work

1. LangChain and LangGraph
2. Claude MCP
3. Awesome-MCP-Servers

Important services

1. MCP - n8n(codeless), Langchain or OpenAI Agents SDK
2. Voice - ElevenLabs, Whisper(Speach to text), xtts2 or Openvoice(Text-to-Speech)
3. Browser Automation - RooCode, browser-use
4. AI Agent (Browser, coding) - RooCode, Screenpipe
5. Internet access - DuckDuckGo, tavily.com
6. Coding - Cursor, windsurf, Replit

Important Links

• OpenAI Whisper TTS example - https://colab.research.google.com/github/fastforwardlabs/whisper-openai/blob/master/WhisperDemo.ipynb

Important theory topics

1. MCP - Model Context Protocol, AI Agents
2. Memory
   • Re-training/Fine tuning
   • RAG - Retrieval-augmented Generation
   • CAG - Cache-Augmented Generation
   • Context Docs
3. AI Workflows
   • Prompt chaining
   • Parllelization
   • Orchestrator-Worker
   • Evaluator-Optimizer
   • Routing

Agents

course

• https://app.pluralsight.com/library/courses/introduction-developing-ai-agents/table-of-contents
• https://github.com/huggingface/agents-course

Useful commands

• List all the fies and their content and save it in a text file find . -type f -exec sh -c 'echo "---Filename: {}---"; cat {}; echo "-------"' \; > files_with_content.txt
• Docker webui for ollama docker run -d --network=host -v open-webui:/app/backend/data -e OLLAMA_BASE_URL=http://127.0.0.1:11434 -p 80:8080 --name open-webui --restart always ghcr.io/open-webui/open-webui:main
• Curl command to talk to DeepSeek API using OpenRouter APIs

  curl https://openrouter.ai/api/v1/chat/completions \
      -H "Content-Type: application/json" \
      -H "Authorization: Bearer sk-or-v1-979c08eebcd708a3da63251099fdbddba75f80690299d25f9d3aexxxxxxx" \
      -d '{
      "model": "deepseek/deepseek-r1:free",
      "messages": [
        {
          "role": "user",
          "content": "What is the meaning of life?"
        }
      ]
      
    }'
  

• Curl command to for free Gemini api key

  curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=YOUR_API_KEY" \
      -H 'Content-Type: application/json' \
      -X POST \
      -d '{
        "contents": [
          {
            "parts": [
              {
                "text": "Explain how AI works in a few words"
              }
            ]
          }
        ]
      }'