🧠 About the Project
🎯 Inspiration
As self-learners, we often struggle with a simple but frustrating question:
"Where do I start, and what should I follow?"
With countless courses, tutorials, and articles scattered online, finding the right path for learning something new is overwhelming.
Course Crafter was born to solve this — an AI tool that designs personalized learning roadmaps in seconds based on your time, budget, and preferences.
🛠 How We Built It
The project was built in a 7–8 hour solo sprint for a hackathon, using:
- 🧠 LLaMA 3 via Modal for large language model inference
- ⚙️ FastAPI to handle backend requests and deploy LLM logic
- 💻 Gradio for building an intuitive and clean frontend UI
- 📄 TXT for generating downloadable course plans
All code was modularized to enable future extensibility, including adding dynamic resource fetching from YouTube, Reddit, and arXiv.
🧩 What I Learned
- How to use Modal's modern deployment flow for serverless AI functions
- Best practices for prompt engineering and formatting output with LLMs
- Integrating Gradio custom components (copy/download buttons, etc.)
- How to streamline user input → AI → actionable output workflow
- Working with formats like
.txtto enhance usability
🚧 Challenges Faced
- Fine-tuning prompts to generate outputs that are useful, structured, and link-rich
- Dealing with asynchronous backend calls to Modal APIs
- Implementing custom download functionality in Gradio beyond default outputs
- Managing file persistence in stateless serverless environments
Overall, Course Crafter was a fast-paced yet rewarding build, showing how powerful AI can be when aligned with real-world learning needs.
Built With
- fastapi
- gradio
- llama3
- modal
- python
Log in or sign up for Devpost to join the conversation.