Skip to main content
Back to challenges
Challenge

Google ADK & NeMo Guardrails Mortgage Prepayment Risk Monitor

AI-driven mortgage refinancing triggers rapid prepayment swings that disrupt mortgage bond valuations. Build a guarded analysis system using Google ADK and NVIDIA NeMo Guardrails to ingest borrower signals, verify policy safety, and predict prepayment speed shifts.

Business OperationsHosted by Vera
Sign in free to start
Challenge brief

What you are building

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

Construct an AI application using Google ADK guarded by NVIDIA NeMo Guardrails to model mortgage prepayment speeds.

Delivery guide

How work is evaluated

Evaluation

Evaluates prepayment rate forecasting precision and adherence to safety guardrails.

Datasets

Challenge datasets

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

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

guardrail_safety_test

Tests that unverified financial advice or illegal statements are blocked by NeMo.

Binary check

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

Dimension 2

cpr_mae

Mean absolute error of predicted Constant Prepayment Rate against benchmarks. • target: 1.2 • range: 0-10

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 a Google ADK (Agent Development Kit) agent connected to Vertex AI / Gemini

  • Define safety and financial advice policies using NeMo Guardrails Colang syntax

  • Analyze rate differential and AI refi tool adoption signals to forecast CPR (Constant Prepayment Rate)

  • Enforce compliance and data privacy guardrails on investment advice outputs

Resources and assets

Reference links and supporting material

Dataset notes

Historical mortgage pool datasets paired with rate shifts and simulated AI-refi platform usage indexes.

Challenge setup

Evaluates prepayment rate forecasting precision and adherence to safety guardrails.

Challenge input

JSON with coupon_rate, current_market_rate, and ai_refi_penetration_index

Google ADK

Provides Gemini agent integrations for financial reasoning.

NVIDIA NeMo Guardrails

Enforces strict financial compliance and safety guardrails.

Evaluated output

JSON with predicted_cpr, risk_tier, and guardrail_passed boolean

Evaluation checks
  • Tests that unverified financial advice or illegal statements are blocked by NeMo.
  • Mean absolute error of predicted Constant Prepayment Rate against benchmarks. • target: 1.2 • range: 0-10
Configured evaluation evidence
  • Benchmark: MortgageRiskBench
  • Cpr Mae target: 1.2
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Google ADK · Required
  • RAI · Optional
  • Google · Optional

Evaluation contract

  • guardrail_safety_test · Weight 1
  • cpr_mae · Weight 1

Metrics

  • cpr_mae · Target 1.2

Benchmarks

  • MortgageRiskBench
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON with coupon_rate, current_market_rate, and ai_refi_penetration_index

    Output: JSON with predicted_cpr, risk_tier, and guardrail_passed boolean

    Execution harness: Python

Evaluation tools

  • Google ADK · Agent Framework
  • NVIDIA NeMo Guardrails · AI Guardrails Tooling

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Google ADK & NeMo Guardrails Mortgage Prepayment Risk Monitor