What I do

Seven kinds of work. One person on the hook.

Each of these is something I've actually shipped, more than once. They overlap on purpose — most engagements blend three or four of them, because real problems don't respect tidy categories.

[01]

Enterprise Architecture & Solution Design

I design the bones of large systems so they don't snap when the business grows. Modular, integrated, secure, and built for the team that has to live with it.

  • Multi-tier architecture that holds up at scale
  • Microservices, modular design, SaaS platforms
  • API-first and event-driven where it actually helps
  • Multi-tenant, role-based platforms done right
  • HA / DR thinking built in, not bolted on
  • Security and compliance baked into the design
  • Database design and data governance from day one
[02]

On-Premise AI & Private LLM Infrastructure

This is the work I'm most excited about right now. Local AI environments where every model, every document, every inference stays inside your perimeter — no cloud APIs, no data leaving the building.

  • Isolated, air-gapped LLM systems end-to-end
  • On-prem GPU and CPU inference setups
  • RAG over your private document sets
  • Fine-tuning open-source models on your data
  • Internal knowledge bases with semantic search
  • Pipelines where nothing escapes the perimeter
  • OCR and document intelligence on your own kit
[03]

Digital Transformation

The phrase is overused. The work isn't. I take operational chaos — the spreadsheets, the manual approvals, the legacy systems people are afraid of — and turn it into something the team can actually run.

  • Workflow digitisation that gets adopted
  • Legacy modernisation without breaking the business
  • Process re-engineering that respects the people doing the work
  • Honest technology assessments and roadmaps
  • Change management that isn't just a deck
  • Capability gap analysis with actual fixes
[04]

Enterprise Integration & Automation

Most of my career has been making systems that hate each other learn to talk. APIs, middleware, government integrations, payments — the unglamorous plumbing that keeps the business honest.

  • API ecosystem design and orchestration
  • Middleware and integration architecture
  • Real-time sync and reconciliation
  • Government systems — IFMIS, eGP, national portals
  • Payment gateway and financial integrations
  • Webhook and event-driven patterns done right
[05]

Full-Stack Technology Leadership

I still write code. That matters because I can validate the architecture myself, own the sensitive parts personally, and call out a bad decision before it ships.

  • Frontend with modern reactive frameworks
  • Backend — Django, REST, GraphQL, the works
  • PostgreSQL performance and schema design
  • PWA and mobile-first architecture
  • Cloud, on-prem, and hybrid deployment
  • CI/CD and the discipline that goes with it
[06]

Project Delivery & Governance

Companies bring me in for this when they've already got the people but no one's running the room. I run the room. Requirements, vendors, risks, go-live — all of it.

  • End-to-end delivery, with my name on it
  • Requirements done properly, not just collected
  • Multi-discipline team coordination
  • Vendor management without the games
  • Risk identification and actual mitigation
  • UAT, go-live, and the messy first month after
[07]

Fractional CTO & Strategic Advisory

Senior tech leadership at the board level, without you having to hire someone full-time. Best for startups, SMEs, and scaling teams that need the wisdom but not the headcount.

  • Board-level technology strategy and roadmap
  • Tech due diligence for investment and M&A
  • Build vs buy vs partner — with reasoning
  • IT governance done by someone who's lived it
  • Team structure and hiring guidance
  • Ongoing CTO leadership on a fractional basis
What I'm most excited about right now

Specialism · Private & Sovereign AI

AI that lives inside your walls, not someone else's data centre.

This is the work I'm most engaged in right now. I design and stand up complete, self-contained AI environments — the model, the documents, the inference, all on your infrastructure. No cloud APIs, no quiet data leaks, no surprise pricing, no third-party watching the queries.

Isolated LLM Deployment

I stand up open-weight foundational models — LLaMA, Mistral, Phi, Gemma, Falcon — on your own GPUs or CPUs, tuned for the workload your team actually runs.

Private RAG

Your policies, contracts, manuals and case histories become a private retrieval layer. The model answers from your truth, with citations — and nothing leaves your perimeter.

Air-Gapped Intelligence

Fully network-isolated AI for defence, government, regulated finance — places where a single outbound packet is a problem. I've built for those rooms.

AI Inside Your Existing Tools

I wire local AI into the workflows you already have — ERP, document management, helpdesks, approval flows — through internal APIs you control.

OCR & Document Intelligence

On-prem pipelines that ingest scans, PDFs and forms, extract what matters, classify it, and route it. Your documents never touch a cloud OCR service.

Fine-Tuning on Your Data

Domain-specific adaptations of open-source models — so the AI actually understands your sector's language, regulations and internal terminology, not just the open internet's.

Reference Design · Interactive

Hover any layer

Organisational Perimeter — Isolated

Staff Interfaces
Web · Desktop · API
Enterprise Apps
ERP · HIMS · Portals
Internal API Gateway
Auth · Rate Limit · Routing

AI Inference Core

Foundational Model
LLaMA · Mistral · Phi · Gemma
Fine-Tuning & Adaptation
Domain-specific models
RAG Engine
Vector · Semantic retrieval
OCR & Document Intelligence
Pipelines & extraction
Internal Document Stores
Policies · Contracts · Manuals
Enterprise Data
Databases · Warehouses

Layer · core

AI Inference Core

LLaMA · Mistral · Phi · Gemma · Falcon

Open-weight foundational models running on your own GPUs or CPUs. You pick the model, you own the weights, and you decide when to upgrade. No quiet model swaps, no surprise pricing.

What's on my desk

The kinds of problems people send me right now.

Not a service menu — just the surface areas I'm currently designing, building, and advising on the most.

01Enterprise Architecture02On-Premise AI Systems03Private LLM Infrastructure04Full-Stack Development05Workflow Automation06API Ecosystems07Intelligent Reporting08Cloud & Deployment Strategy09Operational Platforms10Scalable Enterprise Applications11Data & Process Intelligence12Secure Business Systems13Infrastructure Planning14Enterprise Integrations15Confidential Development & Strategic Oversight

What I build with

The stack I actually reach for.

Open where it matters, sovereign where it counts, and boring enough to still be running in five years.

Architecture

MicroservicesEvent-DrivenAPI-FirstMulti-TenantHA / DRSecurity-by-Design

AI / LLM

LLaMAMistralPhiGemmaFalconRAGFine-TuningOCRNLPVector Search

Backend & Data

DjangoRESTGraphQLPostgreSQLRedisWebhooksSchema Design

Frontend & Mobile

ReactPWAMobile-FirstReactive UI

Integrations

IFMISeGPPayment GatewaysMiddlewareReconciliation

Delivery

CI/CDDevOpsUATHypercareVendor ManagementRisk Control

Let's talk

If you've got a problem that nobody seems to own — that's usually where I come in.

Architecture work, on-prem AI, Fractional CTO, or a transformation programme that needs someone to actually finish it. Send me a note — I read every email.