Challenge

BNPL Credit Fraud Anomaly Detection with Pydantic AI and Fiddler AI

Rapid expansion in buy-now-pay-later services exposes financial networks to sophisticated identity spoofing and synthetic fraud tactics. You will construct a type-safe fraud risk scoring workflow that validates applicant telemetry and flags high-risk transactions with verifiable confidence metrics and zero schema validation errors.

CybersecurityHosted by Vera
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build a type-safe BNPL fraud scoring agent using Pydantic AI and monitor decision explainability with Fiddler AI observability.

Delivery guide

How work is evaluated

Evaluation

Evaluates type enforcement and classification precision for buy-now-pay-later application reviews.

Datasets

Shared data for this challenge

Review public datasets and any private uploads tied to your build.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

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

schema_conformance

Ensures the response validates against BNPLRiskResult Pydantic class

Binary check

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

Dimension 2

precision

Precision of HIGH risk fraud identifications • target: 0.97 • range: 0.92-1

Binary check

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

Learning goals

What you should walk away with

  • Implement structured Pydantic AI agents for type-safe decision pipelines

  • Integrate Fiddler AI model monitoring to trace feature drift and explainability metrics

  • Analyze BNPL checkout telemetry for identity theft and loan stacking signals

  • Achieve zero validation runtime exceptions across dynamic application payloads

Resources and assets

Reference links and supporting material

Dataset notes

2,000 synthetic buy-now-pay-later checkout sessions featuring device intelligence, credit soft-pull data, and fraud labels.

How this agent runs

Evaluates type enforcement and classification precision for buy-now-pay-later application reviews.

Challenge input

JSON payload containing checkout session telemetry

Pydantic AI

Type-safe agent definition and schema validation framework.

Fiddler AI

Provides operational observability and model explainability tracking.

Evaluated output

Validated Pydantic model output containing risk_tier, risk_score, and key_factors

Checks for
  • Ensures the response validates against BNPLRiskResult Pydantic class
  • Precision of HIGH risk fraud identifications • target: 0.97 • range: 0.92-1
Proof of success
  • Benchmark: FraudEval-v2
  • Precision target: 0.97
  • 1 public reference case
Runtime evidence
  • Python execution harness
  • Python sandbox (unavailable on Versalist)
View technical recipe

Configured tools

Action Space
  • Pydantic AI · Required
  • Fiddler AI · Optional
Orchestration
  • Pydantic AI · Required

Evaluation contract

  • schema_conformance · Weight 1
  • precision · Weight 1

Recipe state

This is a preview. The configuration can change before the evaluation recipe is locked.

Frequently Asked Questions about BNPL Credit Fraud Anomaly Detection with Pydantic AI and Fiddler AI