Practical AI, on solid infrastructure.
I help teams put language models to work on real tasks — and run them in a way that is secure, observable and under your control.
What I build
Focused on use cases that save time today, not demos.
LLM integration
Connect models to your existing tools and data through clean APIs, with prompts, tests and fallbacks you can maintain.
Private RAG
Question-answering over your own documents with retrieval, citations and access control — your data stays where you decide.
Agents & workflows
Automations that read, decide and act across email, tickets, files and internal systems, with a human in the loop where it matters.
Document processing
Extraction, classification and summarisation of invoices, contracts and reports into structured data.
Tool access with MCP
Expose internal systems to AI assistants through well-scoped, auditable tool interfaces instead of ad-hoc scripts.
Evaluation & guardrails
Test sets, quality metrics, cost tracking and safety checks, so you know when a change makes things better or worse.
Run it your way
The right hosting depends on your data, budget and latency needs.
Hosted models
Use a commercial model API behind your own gateway: central keys, rate limits, logging and spend control.
Self-hosted models
Open-weight models served on your own hardware when data must not leave your perimeter.
Hybrid
Route sensitive requests to private models and the rest to hosted ones, behind a single interface.
Production, not prototypes
AI features are software. They need the same discipline as everything else.
Have a use case in mind?
Tell me about the task, the data and the constraints — I will tell you honestly whether AI is the right tool.
Get in touch