Inspiration

We wanted to build something simple, fast, and powerful — a tool that shows how accessible AI can be when paired with clean engineering. Our goal was to create a lightweight system that anyone can run locally, understand easily, and extend into a full product. That idea became the foundation for Ignisia Creative Engine.

What it does

Ignisia Creative Engine generates text based on user prompts using a transformer model.
The user enters a prompt in the frontend, the backend processes it through a text‑generation pipeline, and the result appears instantly on the screen.
It’s a minimal but complete AI workflow: UI → API → Model → Response.

How we built it

  • Backend: FastAPI + Uvicorn
  • AI Model: HuggingFace Transformers (GPT‑2)
  • Frontend: HTML + JavaScript
  • Environment: Python virtual environment

Development process

  1. Set up a FastAPI server with a /generate endpoint
  2. Integrated a transformer model for text generation
  3. Built a simple frontend interface to interact with the backend
  4. Connected everything into a smooth, responsive local demo

Challenges we ran into

  • Virtual environment activation issues
  • Missing dependencies (Torch, Transformers)
  • Model loading errors
  • Routing problems when the server was launched from the wrong directory
  • Ensuring the frontend communicated correctly with the backend

Each challenge helped us better understand the full stack and refine the workflow.

What we learned

  • How to integrate transformer models into a FastAPI backend
  • How to debug environment and dependency issues
  • How to build a minimal but functional AI pipeline
  • How to structure a clean UI → API → Model architecture

What's next

  • Improved UI/UX
  • More advanced models
  • Additional AI tasks (summaries, chat, classification)
  • Cloud deployment
  • User accounts and saved generations

Ignisia Creative Engine is just the beginning — a foundation for a more powerful AI platform.

Built With

Share this project:

Updates