Case Studies

What the work actually looks like.

Client names are withheld out of respect for confidentiality; the situations, decisions, and outcomes are described as they happened. Happy to talk through any of these in detail on a call.

Hospitality & Travel · Interim CTO · Current

Agentic AI in production at a luxury destination-club operator

A private destination-club and vacation-rental operator with no in-house technology leadership. The business ran on an aging Rails monolith, pricing data had degraded to the point that it was actively suppressing conversion, and the vendor portfolio had accumulated without review. Technology decisions were being made by whoever happened to be closest to the problem.

The situation

Pricing and availability inconsistent across distribution channels. No trustworthy reporting layer. Integration debt across three separate channel and property-management platforms. Critical system access concentrated in a small number of individuals with no documented custody.

What was done

Took the technology leadership seat reporting to the CEO. Built a modern data lakehouse with dashboards and automated data-quality monitoring, then designed and deployed pricing, data-quality, and conversion agents on AWS Bedrock — in production against live revenue data. Cleaned up channel and PMS integrations, rationalized vendors, and restructured access custody. Ran the modernization of the legacy platform alongside an IT operating-model transition.

Outcomes

  • Agentic AI operating in production, not a pilot
  • Pricing and inventory integrity restored across all channels
  • Single trusted analytics and reporting layer where none existed
  • Key-person and access-custody risk materially reduced
  • An owned, sequenced technology roadmap at the executive table
Financial Services · Principal Enterprise Architect

A dynamic pricing platform that won Engagement of the Year

A financial-sector client needed pricing that responded to market conditions in something closer to real time — without abandoning the governance and auditability their environment required.

The situation

Pricing decisions were slow, backward-looking, and disconnected from the feedback data that would have improved them. Any solution had to be defensible to risk and audit, which ruled out the fast-and-loose approaches on offer.

What was done

Designed a dynamic pricing platform on AWS combining serverless architecture, microservices, MLOps, and real-time feedback analytics — with model lifecycle governance built into the design rather than bolted on after.

Outcomes

  • Recognized with the firm's Engagement of the Year award
  • ML-driven pricing shipped inside a governance-heavy environment
  • Reference architecture reused on subsequent client engagements
Fuels & Energy · Enterprise Architect

~$10M in savings from platform simplification and mainframe decommissioning

A global energy major's downstream business was carrying decades of accumulated platform debt, including mainframe systems that remained in service largely because retiring them looked harder than paying for them.

The situation

An application portfolio no one could fully account for, overlapping capabilities across business units, and legacy platforms whose annual cost was invisible because it had always been there.

What was done

Directed legacy platform modernization and mainframe decommissioning, and designed the transition architecture that supported M&A integration. Rationalized the application portfolio against actual business capability rather than org-chart ownership.

Outcomes

  • Approximately $10M in cost savings enabled
  • Mainframe dependency retired
  • Transition architecture that carried the M&A integration
Fuels & Energy · Data Architecture Lead

100+ analytics use cases delivered on a modern cloud data platform

The same downstream business needed analytics that spanned refining, logistics, sales, and finance — functions that had never shared a data foundation.

The situation

Operational data trapped in function-specific systems. Every cross-functional question required a manual reconciliation exercise, which meant most of them were never asked.

What was done

Deployed Palantir Foundry on AWS and led delivery of more than 100 analytics use cases across the downstream value chain, with automation replacing recurring manual work.

Outcomes

  • 100+ production analytics use cases delivered
  • Measurable FTE cost savings through automation
  • Cross-functional analysis made routine rather than exceptional
Manufacturing · Enterprise Architecture

Untangling the application estate at two large packaging manufacturers

Two global packaging and paper manufacturers, each running dozens of plants assembled over decades of organic growth and acquisition. Neither could answer a basic question with confidence: what applications are we actually running, which ones do the same job, and what would it cost us to stop running the redundant ones?

The situation

Plant-level autonomy had produced parallel systems for the same business capability across sites. Acquisitions arrived carrying their own ERP and their own integrations, and were never fully absorbed. Ownership of the estate was distributed to the point that no single view of it existed.

What was done

Led enterprise architecture and application portfolio rationalization: built a capability-based map of the estate, identified genuine functional overlap as distinct from superficial similarity, and sequenced a retirement and consolidation roadmap against plant operating constraints. Established architecture governance that plant leadership could work with rather than route around.

Outcomes

  • A single capability-based view of the application estate
  • Redundant systems identified and sequenced for retirement
  • Post-acquisition consolidation path defined rather than deferred
  • Architecture governance that held up against plant-level autonomy
Multi-vertical · Practice Leadership

Building a cloud and enterprise architecture practice from the ground up

A global services firm needed a cloud engineering practice that could both sell and deliver — across AWS, Azure, and GCP, for clients in several industries at once.

The situation

Cloud capability existed in pockets but had no coherent go-to-market, no capability model, and no consistent delivery approach across client engagements.

What was done

Led the cloud and enterprise architecture practices end to end: go-to-market strategy, capability development, and client delivery. Served as trusted advisor to client CTOs and CIOs on EA roadmaps, modernization, technical-debt reduction, and agile transformation.

Outcomes

  • Received the President's Award for exceeding revenue and client growth targets
  • Repeatable delivery model across three major cloud platforms
  • Practice positioned as a durable revenue line, not a set of one-off projects

Success patterns

What tends to be true when these engagements work

The data problem comes first

In almost every engagement, the exciting initiative is blocked by an unglamorous data integrity problem. Fixing that first is what makes everything downstream possible — including AI.

AI has to reach production

The difference between an AI strategy and an AI capability is whether something is running against real data with governance around it. Everything else is a slide.

Rationalization pays for the work

Vendor overlap, duplicate capability, and platforms nobody has reviewed in years routinely fund the modernization that follows. The savings are usually already in the building.

Someone has to own the roadmap

A plan without an executive owner reverts to whatever the loudest vendor suggests. Naming that owner is part of the Blueprint, not an afterthought.

Governance enables speed

In regulated environments, the fastest teams are the ones whose controls are designed in. Governance-by-exception is what slows delivery to a halt.

Risk gets found, not reported

Key-person concentration and access custody almost never surface in a status report. They surface when someone deliberately goes looking.

Want the version specific to your business?

The Production AI Blueprint applies the same approach to your systems and gives you a sequenced roadmap, with the governance and controls to execute it, that you own outright.