Week 13–14: Unsupervised Learning Platform — Segmentation, PCA, and FastAPI in One Stack #aiml

This demo walks through the Week 13–14 integrated project: a full unsupervised learning platform built from Days 85–92 of the curriculum.

Instead of running separate scripts for K-Means, optimal cluster selection, hierarchical clustering, and PCA, everything lives in one system with a FastAPI backend, React dashboard, and Docker Compose setup.

In the video, you’ll see:

The Learning Lab exposing lesson concepts over HTTP — K-Means from scratch, sklearn clustering, optimal-K evaluation, PCA theory, and production dimensionality reduction
The product ML layer running real workflows: customer segmentation train/predict, content taxonomy building, and PCA fit with reconstruction scoring
How the architecture separates concerns — learning/ for curriculum-faithful demos, core/ for canonical implementations, and product/ for composed pipelines
The React dashboard for triggering training, inspecting segment profiles, and reading live metrics from the model registry
End-to-end requests via the API — train a segmentation model, predict on new customer records, and verify results in the dashboard
Stack: Python, scikit-learn, FastAPI, PostgreSQL, Redis, React, nginx, Docker.

If you’re moving from ML coursework toward system design, this project shows how unsupervised learning modules become deployable services — not isolated notebooks.

Видео Week 13–14: Unsupervised Learning Platform — Segmentation, PCA, and FastAPI in One Stack #aiml канала Hands On Course Demo