Google's Vertex AI docs tell you which APIs exist; this repository shows how they fit together in working code. It's the official, DevRel-maintained collection of runnable notebooks and sample apps for building generative AI on Google Cloud — and because it tracks Gemini and Vertex features as they ship, the patterns here stay current instead of bit-rotting the way third-party tutorials tend to.
What Sets It Apart
- Maintained alongside the platform. Updated by Google Cloud engineers as new Gemini and Vertex capabilities land, so a notebook you copy today reflects the current API surface rather than a deprecated one.
- Full-stack coverage. Goes beyond basic prompting to function calling, grounding and RAG, embeddings, multimodal input, evaluation, and agent-building frameworks — the pieces you actually assemble for a real application.
- Deployment-minded. Many samples ship with Cloud Run and infrastructure templates, not just isolated snippets, so the jump from notebook to a running service is shorter.
Great Fit If
You're building on Vertex AI or Gemini and want vetted, current reference implementations instead of stitching together blog posts. Look elsewhere if your stack is OpenAI or Anthropic, or if you want a provider-neutral abstraction layer — this is deliberately Google-Cloud-specific and assumes a GCP project, setup, and billing.
