Welcome to the DEMO version of the AI Template Kit for Recommendation Systems — a curated subset of the full production-grade framework designed to help you explore, prototype, and understand how modern recommenders work in practice.
This repository provides a hands-on entry point into the world of collaborative filtering and content-based recommendation, ideal for learners, engineers, and data scientists evaluating the approach before diving deeper.
For the full version (with neural models, APIs, and deployment stack), visit: 🔒 [https://grausoft.net/recommendation-system]
This demo includes two complete learning tracks and introductory previews of the advanced stages:
- ✅ Track 1: Collaborative Filtering (Matrix Factorization with bias terms)
- ✅ Track 2: Content-Based Filtering (using metadata and cosine similarity)
- 🔍 Track 3: Neural Recommenders (intro only – no training code)
- 🔍 Track 4: Operationalization (intro only – no services or deployment code)
Not included in this demo:
- FastAPI inference services
- Dockerized serving infrastructure
- SASRec and Two-Tower training modules
- DOCKER compose setup, model artifacts, or CI/CD setup
This demo uses the full MovieLens 32M dataset and optionally smaller subsets for fast experimentation.
- 👥 200,000+ users
- 🎞 87,000+ movies
- ⭐ 32 million ratings
- 🏷 Metadata for content-based features (genres, tags)
You’ll need to download the dataset yourself. Instructions are included in the intro notebook.
| Folder | Description |
|---|---|
notebooks/ |
Contains Tracks 1 & 2, plus intros for 3 & 4 |
codebase/ |
Shared training and data processing logic |
data/ |
Dataset loading and intermediate storage |
artifacts/ |
Pretrained model weights (for demo runs) |
- Clone the repo and install dependencies
- Start with the onboarding notebook
- Work through
01_collaborative_filteringand02_content_based_filtering - Optionally read the previews of Tracks 3 and 4 to see where this kit can go
Begin your journey with the onboarding notebook:
👉 notebooks/Kit Introduction and Setup Guide.ipynb
It covers:
- How to set up your Python environment
- Where each part of the code lives
- The learning goals and model architecture
- How this demo differs from the full kit
This demo kit is provided for educational and exploratory use. You are welcome to modify, extend, or reference it in your own projects. Please see LICENSE.md for full terms.
Created by Oliver Grau Version: 1.0-Demo | Release Date: 08.07.2025 Website: https://grausoft.net
Let’s build recommendation systems that are modular, understandable, and production-ready — one step at a time.