Skip to content

Getting Started with the Gemini API in Vertex AI with cURL / REST API

A hands-on workshop that walks developers through calling the Gemini API on Vertex AI using cURL and REST: covering text generation, multimodal inputs, function calling, controlled generation, and more.


Type
Workshop
Category
AI
Level
Intermediate
Duration
40 mins
Language
English

Events

Name Organizer Date Location Attendees Links
Build with AI Cloud Manila 2025 GDG Cloud Manila 2025-04-24 Greenfield Tower, Mandaluyong, Philippines 100 📊 Slide Deck
Build with AI Zamboanga 2025 GDG Zamboanga 2025-04-25 Rosal Hall, Universidad de Zamboanga (UZ), Zamboanga City, Philippines 127 📊 Slide Deck
gemini gemini-2.0 vertex-ai generative-ai rest-api curl llm multimodal function-calling api
QR Code for this session
QR Code

Abstract

The full talk abstract: the version submitted to CFPs and shared with event organizers.

In this hands-on workshop, developers will get started with the Gemini API in Vertex AI using nothing but cURL and REST: no SDKs required. Starting from the basics of the Vertex AI API surface, participants will learn how to authenticate, structure requests, and call Gemini models directly via HTTP. Through a series of guided labs, attendees will explore key capabilities including text generation, streaming, multi-turn chat, function calling, multimodal inputs (images and video), controlled generation with JSON schemas, and using Google Search as a grounding tool. By the end of the session, developers will have a working mental model of how the Gemini API works under the hood: knowledge that transfers directly to any language or SDK they choose to use in their projects.

Agenda

  • Introduction: Vertex AI API surface and Gemini model overview
  • Lab 1: Authenticating and making your first API call with cURL
  • Lab 2: Text generation and streaming responses
  • Lab 3: Multi-turn chat conversations
  • Lab 4: Function calling basics
  • Lab 5: Multimodal inputs - sending images and video
  • Lab 6: Controlled generation with JSON schemas
  • Lab 7: Grounding responses with Google Search

Key Takeaways

  • The Gemini API is plain HTTP - no SDK needed to get started, just cURL
  • Understanding the raw API makes you a better developer regardless of which SDK you use later
  • Streaming, multi-turn chat, and function calling are all first-class, built-in API features
  • Multimodal inputs (text, image, video) are supported natively in a single API request
  • Google Search grounding reduces hallucinations by anchoring responses in real-time, factual data