Vibe Coding for AI: Machine Learning to AI Agents

A 2-day hands-on WSQ course covering deep learning with Keras 3 (PyTorch backend), from CNNs to RAG AI Chatbots. All labs use a "vibe coding" workflow where you collaborate with AI assistants to build, debug, and iterate on ML code.

Course Page: Tertiary Courses — WSQ Vibe Coding for AI: Machine Learning to AI Agents

Apply via SkillsFuture: MySkillsFuture Course Directory (TGS-2020503207)

Prerequisites

• Python 3.13+
• uv package manager
• Basic Python programming knowledge
• Familiarity with NumPy and basic ML concepts

Setup

1. Install uv (if not already installed)

# macOS / Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# Windows
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

2. Clone and set up the project

git clone <repo-url>
cd TGS-2020503207-vibe-coding-ml

3. Install dependencies and create virtual environment

uv sync

This automatically creates a .venv with Python 3.13 and installs all dependencies (Keras 3, PyTorch, Gradio, HuggingFace, etc.).

4. Set up environment variables

Create a .env file in the project root with your HuggingFace API token:

HF_TOKEN=your_huggingface_token_here

5. Launch Jupyter

uv run jupyter notebook

Or open notebooks directly in VS Code with the Python extension (select the .venv kernel).

All notebooks use Keras 3 with PyTorch backend. The backend is configured automatically at the top of each notebook.

All notebooks are Google Colab compatible -- just upload and run.

Course Outline — Open in Colab

Click any badge below to open the notebook directly in Google Colab.

Topic 1: Vibe Coding CNNs and AI Agent Image Classifier

CNN MNIST Trainer (Interactive Demo): Launch on Hugging Face Spaces

Lab	Open in Colab
Lab 1: Build a CNN from Scratch (MNIST)
Lab 2: Train CNN with Vibe-Coding (CIFAR-10)
Lab 3: AI Agent Image Classifier

Topic 2: Solving Overfitting Issues with Vibe Coding

Overfitting Explorer (Interactive Demo): Launch on Hugging Face Spaces

Lab	Open in Colab
Lab 1: Diagnose Overfitting (Fashion-MNIST)
Lab 2: Apply Overfitting Solutions
Lab 3: Before/After Comparison

Topic 3: Residual Networks with Vibe Coding

Residual Network Trainer (Interactive Demo): Launch on Hugging Face Spaces

Lab	Open in Colab
Lab 1: Skip Connections & Residual Blocks

Topic 4: HuggingFace Fine Tuning with Vibe Coding

ViT Beans Trainer (Interactive Demo): Launch on Hugging Face Spaces

ViT Beans Classifier (Fine-Tuned Model): Launch on Hugging Face Spaces

Lab	Open in Colab
Lab 1: HuggingFace Fine-Tuning (ViT + Beans)

Topic 5: RAG AI Chatbot with Vibe Coding

RAG AI Chatbot (Interactive Demo): Launch on Hugging Face Spaces

Pocket TTS Demo (Interactive Demo): Launch on Hugging Face Spaces

Lab	Open in Colab
Lab 1: RAG AI Chatbot (ChromaDB + Qwen)