Founder · Engineer · Builder
Regulated industries are deploying AI without the infrastructure to control it. That's the problem I'm solving.
About me
The tools being sold to finance, legal, and healthcare were designed for enterprises where compliance is a feature added at the end.
Not a constraint baked into the architecture from day one. It shows.
What regulated industries actually need isn't a better chatbot.
It's infrastructure that keeps data within their walls, gives them control over what gets queried and by whom.
That's what I'm building.
Before founding The Atlas Company, I spent years across fintech, government, and defence.
I've seen the gap between what AI promises and what organisations in regulated environments can actually ship.
Skills & Expertise
Projects
The AI control layer for regulated industries. Keeps data sovereign, queries governed, and access controlled — built for environments where those guarantees are non-negotiable.
Arbitor's document ingestion layer. Connects to SharePoint, S3, Google Drive, and Azure Blob — chunks, embeds, and indexes into any vector store you control.
The multi-cloud control plane for the Atlas platform. A single interface for provisioning and observing compute, storage, and identity across AWS, Azure, and GCP.
Experience
Led full-stack development of an AI-enabled operational platform for Burger King UK, scaling from 100 to 300+ stores with LLM evaluation pipelines, response caching, and CI/CD-integrated testing.
Delivered enhancements to the BBC's Diversity Tracking platform — automated data publishing workflows, re-architected reporting components, and improved observability across a cloud-native AWS stack.
Delivered AI systems and enterprise platforms across government and defence. Led GenAI adoption organisation-wide, implementing RAG systems, AI chatbots, and cloud-native microservices architectures.
Contracted via Kainos to implement front-end capabilities and backend integrations for a defence platform, delivering against ICD specifications in a containerised, CI/CD-managed environment.
Contracted via Kainos to build and operate IPAFFS — the government's Imports of Plants, Animals, Food and Feed System — across a complex cloud-based microservices environment.
Technical lead for a financial services market data company — responsible for third-party integrations, company app, website, and relational database systems.
Consulted for RedHat at Santander on OpenShift cloud migration. Previously delivered software engineering at American Express.
Writing
Many organisations are no longer struggling to build AI prototypes — they're struggling to scale them. The challenge is rarely the model itself. It's the operational foundations: integration, governance, reliability, and the discipline to move from experimentation to production.
Read on LinkedIn → Agentic AI · AI Infrastructure · StrategyI evaluated Claude and OpenAI Codex against a shared set of engineering tasks on a production-grade open-source codebase — five criteria, five questions, one clear verdict. What emerged wasn't just a ranking, but a clearer picture of how the two tools think differently.
Read more → AI Tools · Engineering · EvaluationBut these aren't AI problems. They are leadership and architecture problems. The organisations that will scale successfully won't be the most experimental — they'll be the most prepared.
Read on LinkedIn → Agentic AI · Governance · ArchitectureBalancing quality with pace means navigating two contending forces — protecting your team from overload while delivering software sustainably. My search for an answer led me to Lean and Kanban, and five practices I now consider essential.
Read on LinkedIn → Engineering · Methodology · LeanContact
If you're responsible for AI adoption in a regulated environment and want to talk about the infrastructure gap — or you're building something in this space — I'd like to hear from you.
kaseosime@btinternet.com