Skip to content

Quickstart

Get the ANZSIC Classifier running on your machine in under 5 minutes.

Prerequisites

Requirement Version Notes
Python 3.12+
PostgreSQL 15 Must have pgvector 0.8+ extension installed
gcloud CLI Latest Must be authenticated (gcloud auth login)
Network — Corporate proxy configured if on the IAG network

1 — Clone and set up the environment

cd /Users/s748779/CEP_AI/anzsic_mapping

# The virtual environment is already created at .venv
# Activate it
source .venv/bin/activate

# Verify the package is importable
python -c "from prod.services.container import get_pipeline; print('OK')"

2 — Configure environment variables

Copy the example env file and fill in your values:

cp prod/.env.example .env

The only values you must change from the defaults:

.env
GCP_PROJECT_ID=top-arc-65ca          # (1)
GCP_LOCATION_ID=australia-southeast1 # (2)
GCLOUD_PATH=/path/to/gcloud          # (3)
HTTPS_PROXY=cloudproxy.auiag.corp:8080  # (4)
  1. Your GCP project ID
  2. The Vertex AI region — must support text-embedding-005 and gemini-2.5-flash
  3. Run which gcloud to find this
  4. Remove this line entirely if you are not behind a corporate proxy

Already configured

If the pipeline was previously working (i.e. the .env file already exists at the project root), you do not need to do anything here.

3 — Verify the database is running

psql dbname=anzsic_db -c "SELECT count(*) FROM anzsic_codes;"

You should see 5236. If the table is empty or missing, run the ingest script:

python ingest.py           # loads anzsic_master.csv → PostgreSQL
python embed.py            # generates embeddings (uses anzsic_embeddings.json checkpoint)

4 — Run your first classification

python -m prod.interfaces.cli --query "mobile mechanic"

Expected output:

────────────────────────────────────────────────────────────
Query : mobile mechanic
Mode  : high_fidelity  |  Candidates: 20
────────────────────────────────────────────────────────────
  #1  [S9419_03] Automotive Repair and Maintenance (own account)
       Class: Other Repair and Maintenance
       Division: Other Services
       Reason: A mobile mechanic who works independently performing
               automotive repair at a customer's location maps directly
               to own-account automotive repair.

python -m prod.interfaces.cli --query "mobile mechanic" --mode fast

streamlit run prod/interfaces/streamlit_app.py

Then open http://localhost:8501 in your browser.

from prod.services.container import get_pipeline
from prod.domain.models import SearchRequest, SearchMode

pipeline = get_pipeline()
response = pipeline.classify(
    SearchRequest(query="mobile mechanic", mode=SearchMode.HIGH_FIDELITY)
)

for result in response.results:
    print(f"#{result.rank}  [{result.anzsic_code}]  {result.anzsic_desc}")
    print(f"      Reason: {result.reason}")

5 — Run the test suite

# Unit + E2E tests (no GCP or DB needed)
python -m pytest prod/tests/unit/ prod/tests/e2e/ -v

# Integration tests (requires live PostgreSQL + GCP auth)
python -m pytest prod/tests/integration/ -m integration -v

Expected: 65 passed for the unit/e2e suite.

Troubleshooting

AuthenticationError: gcloud not found

Set GCLOUD_PATH in your .env file to the full path returned by which gcloud.

DatabaseError: Cannot connect to database

Ensure PostgreSQL is running: brew services start postgresql@15 and that the anzsic_db database exists: createdb anzsic_db

EmbeddingError: Vertex AI Embed returned HTTP 403

Your GCP token has expired or you are not authenticated. Run: gcloud auth application-default login

Proxy / SSL errors

On the IAG corporate network, ensure HTTPS_PROXY=cloudproxy.auiag.corp:8080 is set in your .env file. Do not set HTTPS_PROXY when working off-network.