🌦️ Weather AI Agent

Building AI agents from scratch - No LangChain, no frameworks, just pure Python logic.

🎯 What is this?

An autonomous AI agent that understands natural language and calls tools intelligently to fetch real-time weather data. Built to understand how AI frameworks like LangChain actually work under the hood.

Why I built this:

I was tired of using AI frameworks as black boxes. So I decided to build an agent from first principles to understand:

• How tool calling actually works
• How agents make decisions
• How to parse LLM outputs reliably
• How to build autonomous loops

This is my learning journey in building production-ready AI systems.

---

✨ Features

✅ Autonomous tool calling - Agent decides when to fetch weather
✅ Natural language understanding - Ask in any way you want
✅ Real-time weather data - Uses Open-Meteo API (no API key needed)
✅ Web UI - Clean Gradio interface
✅ Zero frameworks - Pure Python, no LangChain/LlamaIndex

---

🚀 Demo

Try it live: Hugging Face Space

---

🏗️ Architecture

User Query → LLM Decision → Tool Detection → Execute Tool → Format Response

How it works:

1. User asks: "What's the weather in Mumbai?"
2. LLM responds: TOOL_CALLING ...
3. Agent parses and executes tool
4. LLM formats result naturally

No magic. Just clean logic.

---

📦 Installation

# Clone repo
git clone https://github.com/Chebaleomkar/weather-ai-agent.git
cd weather-ai-agent

# Install dependencies
pip install -r requirements.txt

# Run web UI
python app.py

Create api keys at here : https://console.groq.com/keys

Past the keys in .env file

---

💻 Usage

Basic Usage:

from main import AutonomousAgent, LLM

agent = AutonomousAgent(LLM())
response = agent.ask("What's the weather in Paris?")
print(response)

Web UI:

python app.py
# Opens at http://localhost:7860

---

🧠 What I Learned

1. AI frameworks are just abstractions - LangChain, LlamaIndex do exactly what I built here
2. Tool calling is pattern matching - Parse LLM output, execute functions
3. Autonomous agents are loops - Keep going until you get a final answer
4. Prompt engineering matters - How you ask determines what you get

---

🤝 Contributing

This is a learning project, but improvements are welcome! Feel free to:

• Report bugs
• Suggest features
• Submit PRs

---

🙋‍♂️ About Me

I'm Omkar Chebale, learning AI/ML by building from first principles.

• 🔗 LinkedIn
• 🐙 GitHub
• 🐦 Twitter

Currently seeking opportunities in AI/ML Engineering.

---

⭐ If you found this helpful, give it a star!

Building in public. Learning in public. Growing in public.