Skip to main content
Loading navigation
Back to challenges

Automate KYC Intake with Mastra AI and NVIDIA NeMo Guardrails

Automating financial document factories requires strict guardrails to prevent extraction of invalid or hallucinated KYC data. Build a robust document processing pipeline using Mastra AI and NVIDIA NeMo Guardrails that validates unstructured corporate documents.

CybersecurityHosted by Vera
Sign in free to start

What you are building

Implement a TypeScript-based document extraction engine using Mastra AI integrated with NVIDIA NeMo Guardrails to enforce data safety rules.

How work is evaluated

Evaluation

Evaluates accuracy of extracted corporate KYC fields and effective guardrail blocking of corrupted input.

Datasets

Public datasets, plus the private datasets you have access to.

How submissions are scored

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

test_field_accuracy

Verify extracted tax ID format validity

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

extraction_accuracy

Percentage of correctly extracted KYC fields • target: 0.98 • range: 0.9-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

  • Configure Mastra AI agents with structured schema extraction tools.

  • Integrate NVIDIA NeMo Guardrails to reject unsafe or unverified document content.

  • Achieve >95% precision on corporate registration document field parsing.

Resources

Dataset notes

Synthetic unstructured corporate incorporation and tax filings.

Challenge setup

Evaluates accuracy of extracted corporate KYC fields and effective guardrail blocking of corrupted input.

Challenge input

Unstructured text string representing articles of incorporation

Mastra AI

Framework requirement

NVIDIA NeMo Guardrails

AI engineering tool requirement

Evaluated output

JSON containing company_name, tax_id, incorporation_date, registration_status

Evaluation checks
  • Verify extracted tax ID format validity
  • Percentage of correctly extracted KYC fields • target: 0.98 • range: 0.9-1
Configured evaluation evidence
  • Benchmark: Document Parsing Benchmark v1
  • Extraction Accuracy target: 0.98
  • 1 public reference case
Configured runtime
  • JavaScript execution harness
  • javascript sandbox (unavailable on Versalist)

Shows the first task, up to three tools, and up to three evidence items.

View technical recipe

Configured tools

Action Space
  • Mastra AI · Required
Safety / Guardrails
  • NVIDIA NeMo Guardrails · Optional
Orchestration
  • Mastra AI · Required

Evaluation contract

  • test_field_accuracy · Weight 1
  • extraction_accuracy · Weight 1

Metrics

  • extraction_accuracy · Target 0.98

Benchmarks

  • Document Parsing Benchmark v1
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: Unstructured text string representing articles of incorporation

    Output: JSON containing company_name, tax_id, incorporation_date, registration_status

    Execution harness: JavaScript

Evaluation tools

  • Mastra AI · agent_framework
  • NVIDIA NeMo Guardrails · guardrail_tool

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Automate KYC Intake with Mastra AI and NVIDIA NeMo Guardrails