• Create/choose a GCP project and enable Vertex AI API.
• Create a Service Account (for deployments) and set up Application Default Credentials (for local dev).
• Install Google Cloud SDK, authenticate, and verify from a short Python test in PyCharm.

1. Go to Google Cloud Console and sign in with your Google account.
2. Top navbar → Project dropdown → New Project. Name it (e.g., vertex-chat-streamlit).
3. Make sure Billing is enabled for the project (free credits often available for new accounts).

1. Left menu → APIs & Services ▸ Library.
2. Enable Vertex AI API.
3. Also enable: Cloud Storage API (datasets, artifacts), IAM API (permissions).

1. Left menu → IAM & Admin ▸ Service Accounts → Create Service Account.
2. Name: vertex-chatbot-sa.
3. Roles: Vertex AI User. (Later for Cloud Run you’ll also need Service Account User on the runner identity.)
4. Create → open the service account → Keys tab → Add Key ▸ Create new key → JSON → Download.
5. Save as:
   Windows: C:\VertexChatbot\keys\sa.json

   macOS/Linux: /Users/you/VertexChatbot/keys/sa.json

1. Download & install from cloud.google.com/sdk.
2. Open Terminal / CMD / PowerShell and run:

   gcloud init
   gcloud config set project YOUR_PROJECT_ID
   gcloud config set compute/region us-central1
   gcloud auth application-default login

gcloud auth application-default login sets up Application Default Credentials (ADC) on your machine—perfect for PyCharm & Streamlit during development.

CLI checks

gcloud auth list
gcloud config list
gcloud services list --enabled | findstr vertex   # Windows
gcloud services list --enabled | grep vertex      # macOS/Linux

Python check (run in PyCharm)

Install the library in your virtual env:

pip install --upgrade google-cloud-aiplatform streamlit python-dotenv

Create test_vertex.py with:

# test_vertex.py
# Quick smoke test for Vertex AI (Gemini) from your laptop/PyCharm.

import vertexai
from vertexai.generative_models import GenerativeModel

PROJECT_ID = "YOUR_PROJECT_ID"
LOCATION = "us-central1"

vertexai.init(project=PROJECT_ID, location=LOCATION)
model = GenerativeModel("gemini-1.5-pro")  # prebuilt LLM, no training needed
resp = model.generate_content("Say hello in five words.")
print("✅ Call OK. Model replied:")
print(resp.text)

✅ Call OK. Model replied:
Hello there, nice to meet!

Create a .env file at your project root (don’t commit to public repos):

GCP_PROJECT_ID=YOUR_PROJECT_ID
GCP_LOCATION=us-central1
# For key-based auth (optional; prefer ADC in dev)
# GOOGLE_APPLICATION_CREDENTIALS=C:/VertexChatbot/keys/sa.json

Install python-dotenv (already in the earlier pip line) and load in code:

from dotenv import load_dotenv
import os, vertexai
from vertexai.generative_models import GenerativeModel

load_dotenv()
vertexai.init(
  project=os.getenv("GCP_PROJECT_ID"),
  location=os.getenv("GCP_LOCATION","us-central1")
)
model = GenerativeModel("gemini-1.5-pro")
print(model.generate_content("Environment variables loaded successfully.").text)

In Step 2 we’ll create the **Streamlit app shell** (UI), wire it to Vertex AI, and run locally. We will get a chat input box and a response area working end-to-end.