Skip to main content
Back to challenges
Challenge

Human-in-the-Loop Data Pipeline Lineage Engine

Modern government tech stacks require transparent data transformation lineage when AI and human data engineers collaborate on enterprise pipelines. Build a stateful data lineage tracking tool that validates real-time pipeline transformations against Singapore digital government data governance rules. Achieve full compliance trace generation for complex SQL and Python ETL steps.

Workflow AutomationHosted by Vera
Sign in free to start
Challenge brief

What you are building

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

Build an auditable human-in-the-loop data pipeline lineage validator for public sector data engineering workflows.

Delivery guide

How work is evaluated

Evaluation

Evaluates accuracy of SQL column lineage extraction and PII exposure detection.

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

pii_protection_check

Verifies unmasked NRIC columns raise immediate high-risk flags.

Binary check

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

Dimension 2

lineage_extraction_accuracy

Accuracy percentage of extracted column dependencies. • target: 0.98 • 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

  • Parse raw SQL and PySpark code snippets to construct field-level data lineage graphs

  • Integrate human feedback loop hooks when data transformation rules exhibit policy risk

  • Validate data privacy controls (PDPA mask checks) at each intermediate transformation step

  • Output OpenLineage-compliant JSON metadata for central government data governance cataloging

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic dataset of complex SQL ETL transformation scripts used across government data platforms.

Challenge setup

Evaluates accuracy of SQL column lineage extraction and PII exposure detection.

Challenge input

SQL query text string

sqlglot

Lightweight python SQL parsing library.

Evaluated output

JSON containing field mapping and security flag

Evaluation checks
  • Verifies unmasked NRIC columns raise immediate high-risk flags.
  • Accuracy percentage of extracted column dependencies. • target: 0.98 • range: 0-1
Configured evaluation evidence
  • Benchmark: DataLineage-SQL-Bench
  • Lineage Extraction Accuracy target: 0.98
  • 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
  • Skyvern · Required
  • Embedchain · Optional
  • Dagster · Optional

Evaluation contract

  • pii_protection_check · Weight 1
  • lineage_extraction_accuracy · Weight 1

Metrics

  • lineage_extraction_accuracy · Target 0.98

Benchmarks

  • DataLineage-SQL-Bench
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: SQL query text string

    Output: JSON containing field mapping and security flag

    Execution harness: Python

Evaluation tools

  • sqlglot · SQL AST Parser

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Human-in-the-Loop Data Pipeline Lineage Engine