Skip to main content
Loading navigation
Back to challenges

Build a Cross-Border Entity Resolution Workflow with Mastra AI and ZenML

Incomplete entity master data hampers cross-border payment processing and creates regulatory vulnerabilities. Build a Mastra AI agent workflow managed by ZenML pipelines to perform entity resolution, standardize LEI records, and cross-reference ultimate beneficial ownership trees. Deliver a automated entity reconciliation engine achieving greater than 95% precision on dirty enterprise records.

CybersecurityHosted by Vera
Sign in free to start

What you are building

Develop a TypeScript Mastra AI agent framework integrated into a ZenML MLOps pipeline for automated corporate entity data cleaning and disambiguation.

How work is evaluated

Evaluation

Evaluates the agent's ability to resolve ambiguous company names to exact canonical LEI codes.

Datasets

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

How submissions are scored

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

lei_format_check

Checks if output contains a valid 20-character alphanumeric LEI format

Binary check

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

Dimension 2

precision

Precision of resolved LEI matches against ground truth • 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

  • Build a TypeScript Mastra AI agent for structured entity resolution and deduplication

  • Orchestrate training and inference steps using ZenML workflow pipelines

  • Match noisy enterprise trade records against Global Legal Entity Identifier Foundation (GLEIF) databases

  • Construct knowledge graph links for Ultimate Beneficial Owner (UBO) discovery

Resources

Dataset notes

Dataset of 400 noisy cross-border entity records paired with ground-truth GLEIF Legal Entity Identifier records.

Challenge setup

Evaluates the agent's ability to resolve ambiguous company names to exact canonical LEI codes.

Challenge input

JSON object with raw_name, address_snippet, and country

Mastra AI

Offers TypeScript native agent structure and flexible memory integration.

ZenML

Provides structured workflow pipelines for entity data resolution training and inference.

Evaluated output

JSON object with resolved_lei, canonical_name, confidence_score

Evaluation checks
  • Checks if output contains a valid 20-character alphanumeric LEI format
  • Precision of resolved LEI matches against ground truth • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: Entity Resolution Standard Benchmark
  • Precision target: 0.95
  • 1 public reference case
Configured runtime
  • JavaScript execution harness
  • javascript sandbox (unavailable on Versalist)

Shows the first task, up to three tools, and up to three evidence items.

View technical recipe

Configured tools

Action Space
  • Mastra AI · Required
  • ZenML · Optional
Orchestration
  • Mastra AI · Required

Evaluation contract

  • lei_format_check · Weight 1
  • precision · Weight 1

Metrics

  • precision · Target 0.95

Benchmarks

  • Entity Resolution Standard Benchmark
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON object with raw_name, address_snippet, and country

    Output: JSON object with resolved_lei, canonical_name, confidence_score

    Execution harness: JavaScript

Evaluation tools

  • Mastra AI · Agent Framework
  • ZenML · MLOps Pipeline Automation

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build a Cross-Border Entity Resolution Workflow with Mastra AI and ZenML