Practical LLM Applications

This repository is a collection of projects demonstrating practical applications with Large Language Models (LLMs). The key focus is on runimg open-source models locally, enabling development without relying on proprietary APIs.

Each project is contained within its own folder and includes a dedicated README with detailed setup and conceptual explanations.

---

🚀 Projects

Project	Description	Key Concepts Covered
LangChain Conceptual Tutorials	A series of notebooks adapted from official LangChain documentation, progressing from basics to advanced agents.	RAG, Agents, Chatbots, Structured Output, Vector Stores
Analyze Financial Data with AI	"FinVault": A Streamlit application featuring a hierarchical multi-agent team. A "Supervisor" agent delegates tasks to specialized analysts to research stocks and SEC filings.	Multi-Agent Systems, Supervisor-Worker Pattern, Streamlit UI, Tool Calling, Financial Analysis
Chatbot Memory with Ollama & SQLite	"NeuroMind": A full-stack CLI chatbot with persistent memory. Features a FastAPI backend and a beautiful terminal UI. Supports multiple "Personas" and conversation threads.	Client-Server Architecture, SQLite Persistence, FastAPI, Server-Sent Events (SSE), Rich Terminal UI
Long Running Coding Agents	Autonomous Software Engineer: An agent that creates a real workspace, plans features, writes Python code, runs its own tests (pytest), fixes errors, and commits to Git.	Autonomous Agents, Self-Correction Loops, Tool Use (Shell/File System), Plan & Execute
(More projects will be added here...)

---

🛠️ Core Technologies

This repository primarily uses the following technologies:

• [LangChain]: The core framework for orchestrating LLM workflows.
• [Ollama]: For running and managing local LLMs and embedding models.
• [FastAPI]: For building high-performance REST APIs to serve LLM logic.
• [LangGraph]: For building stateful, multi-actor applications.
• [Streamlit]: For building interactive web interfaces for the AI applications.
• [Rich]: For creating beautiful, formatted terminal user interfaces (TUI).
• [Pydantic]: For defining data schemas to enable reliable, structured output.
• [SQLite]: For lightweight, serverless persistent data storage.

🔧 General Setup

1. Clone the Repository:

   git clone https://github.com/majid-200/Practical_LLM_Applications.git
   cd Practical_LLM_Applications

2. Create a Virtual Environment:

   It's highly recommended to use a virtual environment to manage dependencies.

   python -m venv venv
   source venv/bin/activate  # On Windows, use `venv\Scripts\activate`

3. Install Dependencies:

   Install dependencies based on the project you want to run.