Pular para o conteúdo
Venturalítica

Zero to Pro: The 5-Minute Journey

Este conteúdo não está disponível em sua língua ainda.

Goal: Transform from “Python Developer” to “AI Assurance Engineer” in 3 steps.

The Philosophy: Compliance as Code

Section titled “The Philosophy: Compliance as Code”

You are used to pytest for checking if your function adds 2+2 correctly. But how do you test if your AI model respects Human Rights?

Venturalitica treats “Assurance” as a dependency. Instead of vague legal advice, you define stricter Policies (OSCAL). Your CI/CD pipeline enforces them just like linter rules.

The Curriculum

Section titled “The Curriculum”
LevelRoleGoal
Start HereDeveloperRun your first audit in < 60s.
Level 1EngineerImplement Controls for identified Risks.
Level 2IntegratorViz & MLOps: “Compliance as Metadata”.
Level 3AuditorProof: “Trust the Glass Box”.
Level 4ArchitectGenAI Docs: “Annex IV”.

Step 1: Install

Section titled “Step 1: Install”

We recommend uv for blazing speed, or standard pip.

Terminal window
uv pip install venturalitica

Step 2: Get the Code

Section titled “Step 2: Get the Code”

To follow the Academy, clone the samples repository. This will be your working directory for all levels.

Terminal window
git clone https://github.com/venturalitica/venturalitica-sdk-samples.git
cd venturalitica-sdk-samples/scenarios/loan-credit-scoring

Step 3: Run Your First Audit

Section titled “Step 3: Run Your First Audit”

Run this single line of code. It downloads a dataset, loads a policy, and audits a model.

import venturalitica as vl


# Run the 'loan' scenario
vl.quickstart('loan')

Output:

  CONTROL                DESCRIPTION                            ACTUAL     LIMIT      RESULT
  ────────────────────────────────────────────────────────────────────────────────────────────────
  credit-data-imbalance  Data Quality                           0.429      > 0.2      PASS
  credit-data-bias       Disparate impact                       0.818      > 0.8      PASS
  credit-age-disparate   Age disparity                          0.286      > 0.5      FAIL
  ────────────────────────────────────────────────────────────────────────────────────────────────
  Audit Summary: VIOLATION | 2/3 controls passed

Take Home Message

Section titled “Take Home Message”

Step 4: Choose Your Path

Section titled “Step 4: Choose Your Path”

Now that you’ve seen the failure, learn how to fix it and verify it.

  • Level 1: The Engineer — Learn how to implement Controls that mitigate identified Risks. Detect & Block non-compliant models.
  • Level 2: The Integrator — Log outcomes to MLOps tools and verify results visually in the Dashboard.
  • Level 3: The Auditor — Learn how to perform a “Glass Box” audit on the loan model and generate cryptographic proofs.
  • Level 4: The Architect — The Boss Level. Train a high-risk financial model and generate the massive Technical Documentation required by the EU AI Act.

External References

Section titled “External References”