Gemini API with Vertex AI for Developers

Code and slides for the Data For Science webinar series: Gemini API with Vertex AI for Developers (O’Reilly).

This course explores Google’s generative AI stack: the Gemini API (Google AI Studio and Vertex AI) on Google Cloud. You’ll work through text and multimodal use cases, deployment and MLOps patterns on Vertex AI, and end-to-end solution design (RAG, agents, and pipelines). The notebooks are meant to be run in order so concepts and credentials build from one module to the next.

Course structure

1. Gemini API and Vertex AI fundamentals

• Generative AI overview
• Gemini API introduction
• Vertex AI overview

2. Gemini API for text generation

• Prompt engineering for text models
• Text generation with the Gemini API
• Information extraction with text models

3. Multimodal applications with the Gemini API

• Image and video understanding
• Hands-on: captioning and visual Q&A
• Multimodal inputs for richer applications
• Real-world multimodal use cases

4. Vertex AI and deployment strategies

• MLOps practices with Vertex AI (including Kubeflow Pipelines samples)
• Model monitoring and explainability concepts
• Deploying Gemini-powered applications (e.g. Cloud Run-oriented examples)
• Scaling and optimization considerations for production

5. End-to-end generative AI solutions

• Architecture patterns: RAG, agents, and multi-stage pipelines
• Designing production-oriented generative AI systems
• Security, privacy, and responsible use

Setup

Requirements: Python 3.10+ (the materials are tested with 3.12). Dependencies are defined in pyproject.toml and locked with uv.lock.

1) Install dependencies (recommended: uv)

1. Install uv if needed:

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

2. Clone the repo and install into a local virtual environment:

git clone https://github.com/DataForScience/GeminiAI.git
cd GeminiAI
uv sync

uv sync creates .venv and installs Jupyter, Gemini/Vertex client libraries, KFP (Kubeflow Pipelines), and the rest of the stack. To use the environment explicitly:

source .venv/bin/activate   # Windows: .venv\Scripts\activate

2) API keys and Google Cloud

What you need	Used for	Where to get it
GEMINI_API_KEY	Calling the Gemini API from notebooks (Google AI Studio key)	Google AI Studio
GCP_PROJECT	Vertex AI, deployment, and cloud examples	Google Cloud Console (free tier available)
GCP_LOCATION	Vertex AI, deployment, and cloud examples	Google Cloud Console (free tier available)

Set environment variables in your shell or in a .env file in the project root (gitignored). The course uses GEMINI_API_KEY (not GOOGLE_API_KEY).

For Vertex AI and GCP APIs, enable the Vertex AI API on your project and authenticate with Application Default Credentials, for example:

gcloud auth application-default login
gcloud services enable aiplatform.googleapis.com --project=YOUR_PROJECT_ID

Exact gcloud steps also appear in the notebooks where relevant.

3) Launch Jupyter

From the project directory (with the env active, or via uv run):

uv run jupyter notebook

Alternative: pip / venv

This repository is not configured as an installable Python package (no src layout / package discovery), so pip install -e . is not supported. Prefer uv sync.

If you cannot use uv, create a virtual environment and install the dependencies listed under [project].dependencies in pyproject.toml, for example:

python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install "google-cloud-aiplatform>=1.71.1" "google-genai>=1.68.0" "google-generativeai>=0.8.6" "jupyter>=1.1.1" "kfp>=2.13.0" "matplotlib>=3.10.8" "numpy>=2.2.6" "pandas>=2.3.3" "vertexai>=1.71.1" "watermark>=2.6.0"

The checked-in requirements.txt is a minimal, older pin (matplotlib / numpy / pandas / watermark only) and does not include the Gemini, Vertex AI, or KFP stack.

Repository structure

GeminiAI/
├── 1. Gemini Fundamentals.ipynb      # Module 1
├── 2. Text Generation.ipynb          # Module 2
├── 3. Multimodal.ipynb               # Module 3
├── 4. Deployment.ipynb               # Module 4
├── 5. E2E Solutions.ipynb            # Module 5
├── Untitled.ipynb                    # D4Sci notebook template (branding / structure)
├── slides/
│   ├── GeminiAI.pdf                  # Primary slide deck
├── data/                             # Logos, author image, sample images/audio/video
├── d4sci.mplstyle                    # Matplotlib style for figures
├── pyproject.toml                    # Project metadata and dependencies
├── uv.lock                           # Locked versions for uv
├── requirements.txt                  # Legacy minimal deps only (see Setup)
├── LICENSE

Suggested learning path

• Work through Modules 1 → 5 in order.
• For slides only, open slides/GeminiAI.pdf.

Notebook	Topic
1. Gemini Fundamentals.ipynb	Gemini API and Vertex AI fundamentals
2. Text Generation.ipynb	Gemini API for text generation
3. Multimodal.ipynb	Multimodal applications
4. Deployment.ipynb	Vertex AI and deployment strategies
5. E2E Solutions.ipynb	End-to-end generative AI solutions

---

Estimated API costs

Running all notebooks end-to-end typically costs on the order of $1–3 in API usage, depending on how often you re-run cells and current pricing.

Vertex AI / Cloud Run / build steps in Module 4 can incur additional GCP charges if you execute them against a real project.

---

Questions?

Reach out at info@data4sci.com or open an issue if something isn’t working.

Author

Bruno Gonçalves Data For Science, Inc. Web: www.data4sci.com Twitter/X: @bgoncalves LinkedIn: @bmtgoncalves Email: info@data4sci.com Schedule a Call: https://data4sci.com/call