Useful AI,engineered properly.
I help businesses get real value from AI by building systems that fit the way they actually work.
Agents, RAG, workflow automation, data integration, document analysis and product features. Designed, built and shipped as production software, backed by 20+ years of engineering.
Recent work
A few examples of the kind of work behind the words.
Retrieval and reasoning over chemistry data
A retrieval pipeline for flavour and fragrance work, combining chemistry sources and sensory descriptors with model enrichment and comparison.
Read the case study →E-commerce platform architecture
A high-traffic e-commerce rebuild involving legacy URLs, SQL Server, Cloudflare caching, multi-channel storefronts and production migration risk.
Document and policy AI
AI-assisted document validation, approval workflows and SharePoint integration for businesses that need control, auditability and human review.
What I do
Three things I keep building. Different industries, different stacks, same engineering discipline.
Agentic systems
I build AI workflows that can use tools, call APIs, process documents, make decisions and hand work back to people when needed.
The hard part is not the model. It is the guardrails, state, permissions, retries, logging and failure handling around it.
RAG and AI integration
I connect AI to the data and software a business already uses.
That can mean SharePoint, SQL, Postgres, pgvector, Azure AI Search, APIs, document libraries or internal systems. The aim is useful answers and actions, not another chatbot demo.
AI-native product features
I help build AI into products and internal tools from the start.
Search, recommendations, validation, enrichment, workflow assistance and decision support all work better when AI is part of the architecture, not added at the end.
Most AI projects do not fail because the model is weak. They fail because the data is messy, the workflow is unclear, permissions are awkward, nobody owns evaluation, or the demo never had to meet production.
That is where 20+ years of engineering experience helps.
I have worked across software engineering, architecture, consulting, pre-sales and delivery. That means I can talk to users, understand the business problem, design the system, build it, ship it and improve it after real people start using it.
I work best with small and mid-sized businesses that already have systems, data, processes or products — and want AI to improve something real.
That often means specialist service businesses, e-commerce companies, SaaS/product teams, agencies needing a senior technical partner, Microsoft 365-heavy organisations, or businesses with document-heavy workflows.
A practical place to start
AI implementation review
For businesses that know AI could help, but are not sure what to build first.
I review your workflows, data, systems and product ideas, then produce a practical plan: what is worth building, what is not, likely risks, suggested architecture, cost shape and the first useful release.
Fixed scope. Written output. One follow-up call.
Thinking out loud
All writing →Get in touch
Tell me what you're trying to build, or what's not working.