Skip to main content
Back to challenges
Challenge

Build a Mastra AI EPC Milestone Monitor with Braintrust Evaluation for Zuluf Water Expansion

The Saudi Aramco Zuluf water expansion project and Kuwait's massive wastewater deal represent multi-billion dollar infrastructure commitments. This challenge requires developers to build an agentic monitoring system using the Mastra AI framework to orchestrate workflows that track Engineering, Procurement, and Construction (EPC) milestones. You will integrate Braintrust to automate the evaluation of contractor reports against project schedules and regional regulatory standards, specifically focusing on water treatment and desalination localization requirements. The system must ingest unstructured progress reports, extract key performance indicators (KPIs) like 'pipe laying completion percentage' or 'localized procurement spend,' and use Mastra's workflow capabilities to trigger alerts when deviations exceed a 5% threshold. By leveraging Braintrust, you will create a rigorous testing loop that compares the AI agent's extraction accuracy against ground-truth engineering documents, ensuring reliable oversight for GCC megaprojects.

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.

Develop an automated project control workflow using Mastra AI and Braintrust to monitor EPC progress for major water infrastructure projects in Saudi Arabia and Kuwait.

Delivery guide

How work is evaluated

Evaluation

The solution will be evaluated based on the accuracy of KPI extraction and the robustness of the Mastra workflow under failure scenarios.

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

Local Content Validation

Ensures the agent correctly identifies if 'Local Content' targets are met based on the text.

Binary check

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

Dimension 2

Extraction Precision

Percentage of correctly identified numeric milestones • target: 0.95 • 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 complex project management workflows using Mastra AI's state-based transitions

  • Implement automated milestone extraction from semi-structured EPC reports

  • Configure Braintrust for continuous evaluation of agentic reasoning and data extraction

  • Design regional compliance checks for Saudi localized desalination equipment production

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic dataset containing mock EPC progress reports, localization spend logs, and regulatory compliance checklists for Saudi Arabian and Kuwaiti water projects.

Challenge setup

The solution will be evaluated based on the accuracy of KPI extraction and the robustness of the Mastra workflow under failure scenarios.

Challenge input

JSON containing document text and expected schema

Mastra AI

Used for managing stateful AI agent workflows.

Braintrust

Used for logging, tracing, and scoring LLM performance.

Evaluated output

JSON with extracted milestones

Evaluation checks
  • Ensures the agent correctly identifies if 'Local Content' targets are met based on the text.
  • Percentage of correctly identified numeric milestones • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: Infrastructure Monitoring Benchmark
  • Extraction Precision target: 0.95
  • 1 public reference case
Configured runtime
  • Node.js 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
  • Mastra AI · Optional
  • Zed · Optional
Observation
  • Braintrust · Required
Reward / Eval
  • Braintrust · Required
Orchestration
  • Mastra AI · Optional

Evaluation contract

  • Local Content Validation · Weight 1
  • Extraction Precision · Weight 1

Metrics

  • Extraction Precision · Target 0.95

Benchmarks

  • Infrastructure Monitoring Benchmark
  • energy

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing document text and expected schema

    Output: JSON with extracted milestones

    Execution harness: Node.js

Evaluation tools

  • Mastra AI · Orchestration
  • Braintrust · Evaluation

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build a Mastra AI EPC Milestone Monitor with Braintrust Evaluation for Zuluf Water Expansion