Implement Cross-Border Cross-Currency AML Risk Screening using Pydantic AI
International bank expansions require strict cross-jurisdictional compliance screening across BSA and LatAm regulatory frameworks. Develop a type-safe Pydantic AI agent that screens transaction flows against global PEP and sanction lists with 0% false pass rates.
What you are building
Develop a type-safe Python compliance agent with Pydantic AI to enforce multi-jurisdictional AML validation rules.
How work is evaluated
Evaluates false negative rate on international PEP and sanctions list screening.
Datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
Zero False Pass Guarantee
Sanctioned/PEP individuals must never yield pass_flag == True.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Sanctions Match Accuracy
Percentage of accurate sanctions identifications. • target: 1 • range: 0.99-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Learning goals
Build a Pydantic AI agent leveraging strong Pydantic runtime model validation.
Integrate entity resolution tools for cross-referencing US and Brazilian PEP databases.
Enforce strict schema validation on output AML risk classifications.
Ensure zero false negative passes for confirmed sanctions list entities.
Resources
Synthetic cross-border payment stream between US and LatAm banking counterparties.
Challenge setup
Evaluates false negative rate on international PEP and sanctions list screening.
Challenge input
JSON with sender_details, beneficiary_details, origin_country, destination_country
Pydantic AI
Provides type-safe agent execution with Pydantic model validation.
Evaluated output
JSON validated by Pydantic model with pass_flag (boolean), match_details
- Sanctioned/PEP individuals must never yield pass_flag == True.
- Percentage of accurate sanctions identifications. • target: 1 • range: 0.99-1
- Benchmark: CrossBorderAMLBench
- Sanctions Match Accuracy target: 1
- 1 public reference case
- 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
- Pydantic AI · Required
- Pydantic AI · Required
Evaluation contract
- Zero False Pass Guarantee · Weight 1
- Sanctions Match Accuracy · Weight 1
Metrics
- Sanctions Match Accuracy · Target 1
Benchmarks
- CrossBorderAMLBench
- security
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON with sender_details, beneficiary_details, origin_country, destination_country
Output: JSON validated by Pydantic model with pass_flag (boolean), match_details
Execution harness: Python
Evaluation tools
- Pydantic AI · Agent Framework
Recipe freeze marker
No recipe freeze time is recorded.