Selected work
Structured case studies demonstrating problem framing, workflow analysis, governance design, and measurable implementation thinking.
An AI-maturity benchmark of eight European machinery manufacturers, scored only on public disclosure against a rubric fixed before assessment. The finding: across all eight, not one publishes a quantified result from its own AI use.
A full-stack diagnostic platform that scores an 11-question intake across six weighted dimensions using Claude Sonnet, holds every result for human consultant approval before delivery, and offers an optional Stripe-paid deliverable kit.
An interactive business-case dashboard built on public datasets answering where AI adoption investment should go: TAM/SAM/SOM sizing, PESTLE, Porter's Five Forces, ten use cases scored across five dimensions, and a sequenced implementation plan.
A governed, multi-tenant content pipeline for regulated industries, and the reproducible evaluation harness that measures whether the governance holds. Brand-safe guardrails and RAG grounding, scored by three layers: deterministic checks, a context-aware LLM judge on a stronger model than the generator, and an independent ragas cross-check.
A vision-model pipeline that turns a product photo and a catalogue row into a validated listing — and, more to the point, the engineering around the model that decides whether the response is allowed to be stored at all. Strict schema validation, deterministic quality scoring in Python, selective retries, defensive parsing and claim candidates for human review.
Documentation
Selected engagement briefs and supporting project documentation are available upon request at the appropriate stage of a process.
Next step
CV access and LinkedIn provide the career background. Case studies demonstrate structured AI thinking and operating approach.