🛡️ Midinho - A Talking LLM

A fully offline, secure, and open-source chatbot built with:

• 🔧 Ollama (local LLM engine)
• 🧱 Lit (web components for modern UI)
• 🧠 Langchain + Pyodide (LLM orchestration in WebAssembly)
• 🧘 And others
  
  👉 🔗 Live Demo

English Audio Example

Portuguese Audio Example

---

🚀 Project Overview

Private Ollama is a client-side chatbot application that leverages modern web and AI technology to provide:

• ✅ Full local execution of LLM inference
• ✅ Zero data leakage – no internet access required post-load
• ✅ Cross-platform browser support
• ✅ Extendability with WebAssembly-based Python (Pyodide)

---

🧩 Tech Stack Breakdown

Component	Technology	Role
🧠 LLM Engine	Ollama	Serves LLM models locally (e.g., Mistral, LLaMA, etc.)
⚛️ Frontend UI	Lit + Pico CSS	Creates modern, reactive web components
🔗 LLM Pipeline	Langchain + Pyodide	Enables Python-based logic and toolchains in-browser
🌐 Runtime	Pyodide + WebAssembly	Runs Python Langchain in the browser, no server needed

---

🔒 Privacy & Offline Capabilities

🔁 Works Fully Offline

• Once loaded in the browser, no internet is required to:
  • Run the chatbot
  • Query the model
  • Process prompts or tools

🔐 100% Local Data Flow

• No telemetry, logging, or external API calls
• Everything runs:
  • In your browser (UI + logic)
  • On your machine (model served locally via Ollama)

🔋 Use Cases

• Secure research
• Education in remote/offline environments
• Local-only enterprise chatbots
• LLM experimentation sandbox

---

🖼️ App Architecture

+---------------------------+
|        Browser UI         |
| (Lit Web Components)      |
+------------+--------------+
             |
             v
+---------------------------+
|   Pyodide (Python WASM)   |
|  + Langchain Orchestration|
+------------+--------------+
             |
             v
+---------------------------+
|     Ollama Local Engine   |
|   (Runs on localhost:11434)|
+---------------------------+

---

📦 Run on the internet

1. Start Ollama locally on your machine
   Download and install Ollama locally

      # Run llama3
      ollama run llama3
      # Test status
      curl http://localhost:11434
      Ollama is running%

2. 🧪 Open Live Example UI
   👉 https://freedomson.github.io/midinho/

• No backend required
• Loads Pyodide + Langchain in-browser
• Connects to local Ollama for LLM completions

📦 Run on your computer

1. Start Ollama Locally on your machine
   Download and install Ollama locally

      # Run llama3
      ollama run llama3
      # Test status
      curl http://localhost:11434
      Ollama is running%

2. Serve Live LLM Chat Bot with Python

      cd static
      python server.py

3. 🧪 Open Live Example UI
   👉 http://localhost:8000/

📬 Feedback & Contributions

💬 Request features or report issues:
👉 GitHub Issues

📄 License

Private Ollama is open source and respects your digital freedom. Use it. Hack it. Share it.