Skip to main content
Loading navigation
Back to challenges

Build a Cross-Border Cross-Entity AML Monitor with Google ADK and Claude 4 Sonnet

Cross-border expansion by global banking giants requires real-time transaction monitoring across heterogeneous regulatory regimes. Construct an intelligent cross-border AML tracking agent using Google ADK and Claude 4 Sonnet that detects structured money laundering across international corridors.

CybersecurityHosted by Vera
Sign in free to start

What you are building

Build an AML surveillance workflow with Google ADK using Claude 4 Sonnet to identify cross-border structuring and layering activities.

How work is evaluated

Evaluation

Evaluates recall and SAR accuracy for cross-border structuring detection.

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_sar_flagging

Verify true positive structuring triggers flag_sar = true

Binary check

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

Dimension 2

structuring_recall

Recall metric on synthetic structuring datasets • target: 0.97 • 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

  • Implement multi-jurisdiction transaction parsing using Google ADK abstractions.

  • Leverage Claude 4 Sonnet for advanced pattern recognition across complex wire transfer sequences.

  • Detect smurfing/structuring threshold bypasses with high precision.

Resources

Dataset notes

Synthetic international wire transactions connecting Brazilian and US bank accounts.

Challenge setup

Evaluates recall and SAR accuracy for cross-border structuring detection.

Challenge input

JSON array of wire transfers over a 48-hour period

Google ADK

Framework requirement

Claude 4 Sonnet

Model requirement

Evaluated output

JSON with flag_sar, detected_pattern, and total_structured_amount

Evaluation checks
  • Verify true positive structuring triggers flag_sar = true
  • Recall metric on synthetic structuring datasets • target: 0.97 • range: 0.9-1
Configured evaluation evidence
  • Benchmark: AML Wire Benchmark v1
  • Structuring Recall target: 0.97
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python 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
  • Google · Required
  • Google ADK · Optional
Policy Serving
  • Claude 4 Sonnet · Optional

Evaluation contract

  • test_sar_flagging · Weight 1
  • structuring_recall · Weight 1

Metrics

  • structuring_recall · Target 0.97

Benchmarks

  • AML Wire Benchmark v1
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON array of wire transfers over a 48-hour period

    Output: JSON with flag_sar, detected_pattern, and total_structured_amount

    Execution harness: Python

Evaluation tools

  • Google ADK · agent_framework
  • Claude 4 Sonnet · llm

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build a Cross-Border Cross-Entity AML Monitor with Google ADK and Claude 4 Sonnet