Skip to main content
Back to challenges
Challenge

Build a Parametric River Risk Multi-Agent Crew with CrewAI and Qwen 3

In light of severe low-water shipping disruptions on major industrial inland waterways like the Rhine River, insurers require real-time risk assessment and automated parametric insurance claim triggers. In this challenge, you will orchestrate a multi-agent workflow using CrewAI to analyze hydrological gauge sensor feeds, freight transport exposure data, and weather forecasts to assess cargo delay risk and compute parametric payout eligibility. Using Qwen 3 as the primary LLM reasoning backbone, you will configure specialized CrewAI agents including a Hydrology Sensor Analyst, Freight Vulnerability Evaluator, and Parametric Settlement Auditor. The agents must collaborate asynchronously, process time-series river levels, and produce fully audited, machine-readable trigger determinations with calibrated uncertainty estimates. Your completed system will ingest simulated sensor streams, validate cross-source evidence, evaluate contractual parametric thresholds (e.g., Kaub gauge depth falling below 78cm for 48 consecutive hours), and calculate claim payouts according to predefined policy schedules.

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.

Orchestrate specialized CrewAI agents powered by Qwen 3 to analyze real-time hydrological data and automate parametric cargo insurance claim triggers.

Delivery guide

How work is evaluated

Evaluation

Evaluates the agentic workflow execution, metric accuracy, parametric trigger determination precision, and JSON schema output compliance.

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

json_schema_validation

Ensures the agent output adheres to the required JSON output schema.

Binary check

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

Dimension 2

parametric_trigger_accuracy

Ensures the payout decision matches ground truth boolean logic.

Binary check

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

Dimension 3

payout_accuracy

Precision of calculated insurance payout relative to contract guidelines. • target: 1 • range: 0-1

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

  • Orchestrate collaborative multi-agent workflows using CrewAI

  • Integrate Qwen 3 for structured hydrology and financial reasoning

  • Process time-series river level feeds to evaluate parametric trigger rules

  • Generate verifiable settlement audit reports with explainable evidence

Resources and assets

Reference links and supporting material

Dataset notes

Simulated Rhine River telemetry data containing hourly water gauge depth levels across key locations (Kaub, Duisburg, Cologne) paired with synthetic parametric insurance contract policies.

Challenge setup

Evaluates the agentic workflow execution, metric accuracy, parametric trigger determination precision, and JSON schema output compliance.

Challenge input

JSON containing river gauge time-series readings, contract parameters, and policy details

CrewAI

Provides structured multi-agent orchestration with role assignment and tool delegation.

Qwen 3

Offers strong technical reasoning and structured JSON output generation for domain workflows.

Evaluated output

JSON containing claim_triggered (bool), payout_amount (float), gauge_station (str), and evidence_summary (str)

Evaluation checks
  • Ensures the agent output adheres to the required JSON output schema.
  • Ensures the payout decision matches ground truth boolean logic.
  • Precision of calculated insurance payout relative to contract guidelines. • target: 1 • range: 0-1
Configured evaluation evidence
  • Benchmark: ClimateRiskBench-InlandWaterways
  • Payout Accuracy target: 1
  • 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
  • CrewAI · Optional
  • Zed · Optional
Policy Serving
  • Qwen 3 · Required

Evaluation contract

  • json_schema_validation · Weight 1
  • parametric_trigger_accuracy · Weight 1
  • payout_accuracy · Weight 1

Metrics

  • payout_accuracy · Target 1

Benchmarks

  • ClimateRiskBench-InlandWaterways
  • energy

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing river gauge time-series readings, contract parameters, and policy details

    Output: JSON containing claim_triggered (bool), payout_amount (float), gauge_station (str), and evidence_summary (str)

    Execution harness: Python

Evaluation tools

  • CrewAI · agent_framework
  • Qwen 3 · llm

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build a Parametric River Risk Multi-Agent Crew with CrewAI and Qwen 3