GW
HomeWorkAboutNotesStackLab
Experience atOmnicom Group

Driving financial systems and automation across global teams.

Builder ofEnterprise AI Systems

AI agents, audit workflows, and decision intelligence at scale.

Based inNew York City

Building systems for enterprises around the world.

HomeNotesWorkAboutStackLabPartners
XLinkedInGitHubEmailllms.txt
© 2026 · New York City
All enterprise work

Controls & assurance case study

ESG reporting and assurance pipeline.

A cross-platform data and control process for producing ESG reporting inputs, improvement signals, and limited-assurance evidence.

Microsoft Azure logoAzureMicrosoft Dynamics 365 logoDynamics 365Workday logoWorkdaySAP logoSAPOracle logoHyperionAribaDatabricks logoDatabricks
Ask the project copilotGrounded in this public-safe case study.

Business problem

What people needed to solve.

ESG disclosures pull from operational, finance, workforce, and vendor systems, and each reported figure has to trace back to a source and a method for internal review and limited assurance. Without that lineage, assurance teams cannot answer where a number came from or how it was derived.

Decision frame

Questions the work needed to answer.

  • Which sources contribute to each reporting output?
  • Where are the quality gaps, anomalies, or opportunities for improvement?
  • What evidence is available for internal review and limited assurance?
Interactive operating flowSelect any node for context or ask the AI guide.

Selected Operating layer

Azure quality workflows + Databricks

Designed an Azure-based ingestion and data-quality workflow that brings approved operational, vendor, and finance data into curated reporting views, with an AI-assisted analysis layer in Databricks for anomaly review and improvement signals. Reporting views retain source, transformation, and review context, and third-party reporting keeps defined evidence, ownership, and review cycles.

Implementation

From the business problem to a working operating model.

Designed an Azure-based ingestion and data-quality workflow that brings approved operational, vendor, and finance data into curated reporting views, with an AI-assisted analysis layer in Databricks for anomaly review and improvement signals. Reporting views retain source, transformation, and review context, and third-party reporting keeps defined evidence, ownership, and review cycles.

  1. 01

    Ingested approved data from ERP, HCM, planning, procurement, and vendor sources.

  2. 02

    Applied multiple quality and transformation workflows before creating reporting views.

  3. 03

    Created a Databricks analysis layer for anomaly review and improvement signals.

  4. 04

    Produced structured reporting files for third-party footprint reporting and assurance workflows.

Controls, approvals, and delivery constraints

The work around the work.

Enterprise systems change only when data, access, testing, ownership, and evidence move together.

  • Reporting views retain source, transformation, and review context.
  • Third-party reporting and limited assurance require defined evidence, ownership, and review cycles.
  • No emissions figure is shown publicly without an approved disclosure source.

Prior art

How leading teams approach this.

Public references for the same class of problem, so this work can be read against how other organizations are building it.

Watershed

CSRD software for reporting and assurance

Traces every reported metric back to source data, calculation, and approver, with role-based ownership, approval chains, automated checks, and direct auditor access for limited assurance review.

Persefoni

Preparing for GHG assurance: a guide to audit-ready emissions reporting

Frames assurance readiness as process maturity: activity data in a controlled environment, source documentation tied to each data point, and documented methodology, assumptions, and boundary decisions.

Public-safe outcome

A controlled route from multi-system inputs to reporting-ready views and assurance evidence, with no emissions figure shown publicly unless it comes from an approved disclosure source.