AI Toolbox Experiments

A collection of practical AI experiments and tools exploring various frameworks and libraries for building intelligent applications.

🛠️ Projects

AutoGen Multi-Agent Development

autogen-dev-team/
Simulates a mini development team using Microsoft's AutoGen framework. Two AI agents collaborate: one generates Python sorting algorithms while another tests and critiques the code through iterative conversations.

FastAPI + OpenAI Code Refactoring

fastapi-openai-refactor/
A REST API service that leverages OpenAI's models to automatically refactor code for better efficiency and readability. Submit code via /refactor endpoint and receive improved versions with explanations.

Gradio Sentiment Analysis UI

gradio-sentiment-ui/
A web-based sentiment analysis tool built with Gradio and Hugging Face transformers. Analyze text emotions through a simple interface with downloadable CSV results.

Haystack Document Q&A

haystack-doc-qa/
Intelligent document search using Haystack's NLP framework. Indexes technical blog PDFs and enables natural language queries with sourced answers from the documents.

LangChain RAG Agent

langchain-rag-agent/
Retrieval-Augmented Generation (RAG) implementation using LangChain for building context-aware AI agents.

Pandas AI Data Analysis

pandas-ai-queries/
Natural language data analysis using PandasAI. Query datasets in plain English and get automated insights and visualizations.

Whisper Podcast Processor

whisper-pydub-summarizer/
Audio processing pipeline that segments podcasts, transcribes content using OpenAI Whisper, and generates summaries with action items using GPT models.

📂 Readme(s)

Adapted from GPT

🚀 Getting Started

Each project contains its own README with detailed setup instructions and usage examples. Most experiments require:

• Python 3.10+
• OpenAI API key (for GPT-powered features)
• Project-specific dependencies (see individual requirements.txt files)

📋 Requirements

Individual projects may require additional setup:

• Audio processing: FFmpeg installation
• PDF processing: PyPDF libraries
• Web interfaces: FastAPI, Gradio, or Streamlit
• Vector databases: FAISS, Elasticsearch (for advanced search)

🎯 Purpose

These experiments demonstrate practical applications of:

• Multi-agent AI systems
• Natural language processing
• Document intelligence
• Audio/video processing
• Web API development
• Interactive ML interfaces