Alright, here's a professional-grade folder structure + code scaffolding plan for the AI Knowledge Engine, designed to keep it simple, clean, and deployable within 10 hours.
knowledge_engine/
├── backend/
│ ├── manage.py
│ ├── requirements.txt
│ ├── backend/ # Django project folder
│ │ ├── __init__.py
│ │ ├── settings.py
│ │ ├── urls.py
│ │ ├── wsgi.py
│ ├── api/ # Django app
│ │ ├── __init__.py
│ │ ├── views.py
│ │ ├── urls.py
│ │ ├── embeddings.py # Embedding + retrieval logic
│ │ ├── llm_response.py # Call HuggingFace / LLM for final answer
│ │ ├── sample_data.txt # Preloaded knowledge base (sample text)
├── frontend/
│ ├── index.html
│ ├── assets/
│ │ ├── styles.css
│ │ ├── script.js
├── README.md
-
backend/settings.py
-
Setup basic Django config + CORS allow all origins
-
Add rest_framework, corsheaders, api to INSTALLED_APPS
-
-
backend/urls.py
- Route all API endpoints (e.g., /api/ask)
-
api/urls.py
- Define path('ask/', AskView.as_view())
-
api/views.py
-
Class AskView(APIView):
-
POST method: receives question
-
Calls retrieval function
-
Calls LLM response generation
-
Returns final answer as JSON
-
-
-
api/embeddings.py
-
Load sample_data.txt
-
Embed text chunks
-
Implement basic semantic search (cosine similarity or FAISS)
-
-
api/llm_response.py
- Use HuggingFace pipeline (or API) to generate answer from matched context.
-
Djangodjangorestframeworktransformerssentence-transformerstorchfaiss-cpudjango-cors-headers
-
index.html
-
Simple Bootstrap layout
-
Header
-
Input field for question
-
"Ask" button
-
Answer card
-
-
-
assets/styles.css
- Custom minor styling (loading spinner, answer card formatting)
-
assets/script.js
-
Fetch question via POST /api/ask
-
Show loading spinner
-
Render the answer dynamically on the page
-
-
Add gunicorn to requirements.txt for production
-
web: gunicorn backend.wsgi
-
Set environment variables like DJANGO_ALLOWED_HOSTS=*
-
Frontend can be pushed as a static site if needed, or served via Django staticfiles.
Time SlotDeliverable1st HourDjango + DRF skeleton setup2nd-3rd HourBuild embedding + retrieval modules4th HourSetup API + Test with Postman5th HourFrontend HTML + Bootstrap UI6th HourJS fetch integration7th-8th HourFinal polish: loaders, error handling9th HourDeployment setup10th HourLive Demo + GitHub push
✅ Working full-stack AI Knowledge Engine✅ Hosted live version (Render or HuggingFace Spaces)✅ GitHub repo with clean README
Would you also like me to give you a starter views.py and script.js template?(Literally plug-and-play in class.) 🚀Ready if you want it!