AutoSAS 🚀

AutoSAS is an AI-powered prototype that generates complete websites from natural language prompts.
The system asks the user to describe their website idea and automatically produces the corresponding HTML/CSS/JavaScript files, packaged and ready to download.

Built during a hackathon challenge, AutoSAS demonstrates how large language models (LLMs) can be orchestrated in a multi-agent workflow to automate software creation beyond boilerplate code.

🔑 Key Features

• Natural Language to Code – Users describe a website, and AutoSAS generates a functional template.
• Multi-Agent LLM Pipeline –
  • Developer Agent: generates draft website code (HTML/CSS/JS).
  • Debugger Agent: analyzes errors and regenerates corrected code.
• Automated Error Handling – Code is compiled, diagnosed, and iteratively refined for robustness.
• Extensible Architecture – While demonstrated for websites, the pipeline can generalize to other software domains.
• Hackathon-Built – Created in a fast-paced team setting, where I served as tech lead.

⭐️ Why It Matters

Traditional website development starts with repetitive setup work. AutoSAS showcases how AI-assisted coding can:

• Accelerate development by automating scaffolding.
• Improve reliability with iterative debugging loops.
• Serve as a proof of concept for AI-driven SaaS applications.

This project highlights skills in generative AI, prompt engineering, LLM orchestration, error diagnosis, and full-stack development - areas with direct relevance to modern ML/AI engineering roles.

⚙️ Setup

This project requires the following Python packages:

• Flask

Install dependencies using:

pip install -r requirements.txt

▶️ Usage

1. Add your OpenAI key as an environment variable:

   • macOS/Linux:

     export OPENAI_API_KEY="your_api_key_here"

   • Windows (PowerShell):

     setx OPENAI_API_KEY "your_api_key_here"

2. Run the app:

   python main.py

3. Open the provided local URL in your browser.

4. Enter a description of your website idea → receive a downloadable .zip file with the generated code.

🛠️ Technology

• Python Flask backend for UI and API handling.
• OpenAI API for LLM-driven code generation and debugging.
• Multi-step pipeline design for reliability and extensibility.

📌 Future Directions

• Expand beyond websites to generate end-to-end SaaS applications.
• Integrate deployment workflows for one-click publishing.
• Explore reinforcement mechanisms for improving LLM code quality.