Skip to main content
Back to challenges
Challenge

Nuclear Verdict Litigation Risk Forecaster with Google ADK and Daytona

The 40.7% surge in nuclear jury verdicts demands proactive legal risk prediction during casualty claims processing. You will develop a distributed legal analytics system using Google ADK and Daytona sandbox environments to process litigation transcripts and predict liability explosion risks. Achieve an AUC-ROC score of at least 0.87 on casualty verdict prediction.

Machine LearningHosted by Vera
Sign in free to start
Challenge brief

What you are building

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

Deploy Google Agent Development Kit within Daytona containerized sandboxes to predict nuclear legal verdict hazards.

Delivery guide

How work is evaluated

Evaluation

Evaluate verdict size prediction accuracy and nuclear risk probability calibration.

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

probability_bounded

Risk probability must be between 0.0 and 1.0 inclusive.

Binary check

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

Dimension 2

auc_roc

Area Under ROC Curve for nuclear verdict prediction • target: 0.87 • range: 0.5-1

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

  • Develop multi-agent workflows using Google ADK (Agent Development Kit)

  • Isolate legal code execution and model pipelines in Daytona sandboxes

  • Extract litigation features from courtroom transcript feeds

  • Predict high-severity jury payout probabilities using ensemble techniques

Resources and assets

Reference links and supporting material

Dataset notes

Historical civil court verdict transcripts, venue risk indices, and synthetic case briefs.

Challenge setup

Evaluate verdict size prediction accuracy and nuclear risk probability calibration.

Challenge input

JSON containing case details, venue, plaintiff claims, and injury severity

Google ADK

Structured agent coordination and Gemini API integration.

Daytona

Isolated development environment for safe multi-agent execution.

Evaluated output

JSON with nuclear risk probability and risk factor breakdown

Evaluation checks
  • Risk probability must be between 0.0 and 1.0 inclusive.
  • Area Under ROC Curve for nuclear verdict prediction • target: 0.87 • range: 0.5-1
Configured evaluation evidence
  • Benchmark: LegalRiskBench-2026
  • Auc Roc target: 0.87
  • 1 public reference case
Configured runtime
  • Docker 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

Environment
  • Daytona · Optional
Action Space
  • Google · Required
  • Daytona · Optional
  • BentoML · Optional

Evaluation contract

  • probability_bounded · Weight 1
  • auc_roc · Weight 1

Metrics

  • auc_roc · Target 0.87

Benchmarks

  • LegalRiskBench-2026
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing case details, venue, plaintiff claims, and injury severity

    Output: JSON with nuclear risk probability and risk factor breakdown

    Execution harness: Docker

Evaluation tools

  • Google ADK · Agent Framework
  • Daytona · Sandbox Runtime

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Nuclear Verdict Litigation Risk Forecaster with Google ADK and Daytona