Skip to main content
Back to challenges
Challenge

LlamaIndex Construction Risk: Underwriting Knowledge Transfer Engine

Brokerage talent moves often result in unquantified portfolio risk leakage and unstructured document disorganization during team transitions. Build a Groq-accelerated LlamaIndex RAG pipeline to ingest construction underwriting guidelines and extract policy transfer exposure metrics within 10 seconds.

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.

Implement a high-speed LlamaIndex retrieval pipeline hosted on Groq Cloud to analyze commercial construction policy portfolios during broker transitions.

Delivery guide

How work is evaluated

Evaluation

Evaluates RAG retrieval precision and extraction speed on commercial policy documents.

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

latency_under_10s

Validates execution completes within 10 seconds.

Binary check

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

Dimension 2

exposure_accuracy

Accuracy of total portfolio monetary exposure calculation. • target: 0.96 • 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

  • Build a LlamaIndex VectorStoreIndex over large construction policy PDFs.

  • Integrate Groq Cloud low-latency LPU endpoints for LLM inference.

  • Extract transferred client exposure values and policy renewal dates.

  • Achieve end-to-end extraction query latency under 10 seconds per portfolio.

Resources and assets

Reference links and supporting material

Dataset notes

Collection of 50 multi-page commercial construction policy documents and underwriting notes.

Challenge setup

Evaluates RAG retrieval precision and extraction speed on commercial policy documents.

Challenge input

PDF document texts representing construction policies

LlamaIndex

Data framework for structured document retrieval and indexing.

Groq Cloud

Provides low-latency LLM execution via LPU hardware.

Evaluated output

JSON containing aggregate_exposure_usd, total_accounts, high_risk_accounts

Evaluation checks
  • Validates execution completes within 10 seconds.
  • Accuracy of total portfolio monetary exposure calculation. • target: 0.96 • range: 0-1
Configured evaluation evidence
  • Benchmark: Enterprise Document RAG Benchmark
  • Exposure Accuracy target: 0.96
  • 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
  • Llama Index · Required
  • Nvidia · Optional
Policy Serving
  • Groq Cloud · Optional

Evaluation contract

  • latency_under_10s · Weight 1
  • exposure_accuracy · Weight 1

Metrics

  • exposure_accuracy · Target 0.96

Benchmarks

  • Enterprise Document RAG Benchmark
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: PDF document texts representing construction policies

    Output: JSON containing aggregate_exposure_usd, total_accounts, high_risk_accounts

    Execution harness: Python

Evaluation tools

  • LlamaIndex · RAG Framework
  • Groq Cloud · Inference Acceleration

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about LlamaIndex Construction Risk: Underwriting Knowledge Transfer Engine