5 min read SDK v0.9.x Python 3.8+

Quickstart

This guide walks you from zero to your first Inferpathio governance event in under 5 minutes. Prerequisites: Python 3.8+, an Inferpathio account, and an existing model you want to govern.

1. Install the SDK

terminal 
pip install inferpathio

2. Get your API key

Navigate to Settings → API Keys in the Inferpathio dashboard. Create a new key with write scope. Store it as an environment variable:

terminal 
export INFERPATH_KEY="ifp_live_xxxx..."

3. Initialize the SDK

init.py 
import os
import inferpathio as ifp

ifp.init(api_key=os.getenv('INFERPATH_KEY'))
# SDK is now initialized for this process

4. Attach governance to your first model

Wrap your existing model logging code with a governance_run context manager:

train.py 
with ifp.governance_run(
    model_id='my-first-model'
) as run:
    # Your existing model logging code
    run.log_model(model, flavor='sklearn')
    run.set_policy('default')
    run.tag('env', 'staging')
    run.tag('team', 'risk-analytics')

5. Log prediction batches for drift monitoring

predict.py 
tracker = ifp.ModelTracker(
    model_id='my-first-model',
    policy='default'
)

# Call after each prediction batch
tracker.log_prediction_batch(
    features=X_batch,
    labels=y_actual,
    threshold=0.1  # PSI threshold for drift alert
)

Next steps