Services
Ways to work with me
Every engagement starts the same way: establish what the problem costs, then decide whether building anything is justified. Most work begins with a diagnostic and moves into a build once the value case holds.
Short demonstration · 1:22
See DomainOS Managed Inference in action
DomainOS — Productised services
Own the intelligence your AI runs on — and the model behind it.
Two services under DomainOS: a Domain Intelligence Layer that turns your knowledge and judgment into assets you control, and a private open-weight model API deployed in your own cloud and run for you.
Build and own the intelligence of your business.
I define and build your Domain Intelligence Layer: the knowledge, decision logic, tools, authority and operating experience that make AI useful in your business — and stay under your control as models change.
Domain representationAuthoritative knowledgeDecision logic and evaluationsTools and authorityInstitutional memoryA learning processDefine your intelligence layer 02DomainOS · Managed InferenceYour own private model API, in your cloud, run for you.
I deploy a licensed open-weight model in your own cloud account, deliver a secured, tested API endpoint, then manage it. You keep your data, and your compute bill stays with your provider.
Qwen · Llama · Mistral · GemmaSee Managed Inference Consulting engagements
Work with me directly
Optimization & Reliability
For existing automation or AI infrastructure that needs to become dependable.
- Monitoring
- Failure handling
- Integration maintenance
- Performance
- Cost optimization
- Additional controls
Domain Model Fine-Tuning & Evaluation
Adaptation only after prompting, retrieval and tool use have been ruled out on the evidence.
- Dataset preparation
- Baseline
- SFT · LoRA / QLoRA
- Embedding & reranker tuning
- Failure analysis
- Model card
- Monitoring & rollback
How an engagement starts
Discovery call
30 minutes on the workflow, what it costs and whether I'm the right person for it.
Diagnostic
Seven days to quantify the problem and design the target architecture.
Build decision
A value case with numbers. Sometimes the answer is that building isn't worth it.
Sprint or retainer
Build the system, or take ongoing ownership of the roadmap and architecture.
Discovery questionnaires
Or start in writing
These cover the same ground as the discovery call, in more detail and in your own time. One question per screen, sections that do not apply are skipped, and everything saves as you go — so you can stop and come back to it. Answers come straight to me.
AI Engineering Discovery
Discovery for LLM, RAG, agentic and production AI engagements: current workflow, the AI / non-AI boundary, evaluation, permissions and reliability.
~25 min · 86 questions →
02Business Workflow Automation Discovery
Map one business process end to end — trigger, steps, systems, exceptions, approvals and cost — and find the parts worth automating.
~20 min · 82 questions →
03Finance Automation Discovery
Map a finance process end to end — source of truth, matching logic, approvals, segregation of duties, exceptions and audit trail — before any of it is automated.
~30 min · 136 questions →