Systems, not slides.
Ten years installing delivery spines on enterprise programs. Now applying that discipline to AI in production for US tech.
I started in IT infrastructure, which is why I still think in systems, not slides. After a few years in the field I realized the problems worth solving weren't technical. They were organizational.
I spent years in consulting building governance frameworks and dashboards, and eventually understood: no dashboard ever fixed a people problem. That pushed me into organizational culture research, deep enough to write a dissertation on it. What I came out with was a practical diagnostic toolset I've been applying in the field ever since.
When AI arrived, my inner sysadmin woke up. Not to chase the hype. To figure out what actually holds in production. I've since built my own workflows: a meeting-to-summary pipeline that cuts post-call overhead to minutes, a Jira/Confluence chain that manages my own outreach cadence and surfaces overdue actions. Nothing I'd bring to a client I haven't run myself first.
Ten years on large enterprise programs across regulated pharma, utilities, energy, and FMCG. SAP, ServiceNow, SAFe. The kind of programs where the gap between "initiative approved" and "in production" is measured in years, not quarters.
What I've learned is that broken programs almost always lose the plot in the same three places: ownership, cadence, and whether AI actually lives in the workflow or just next to it. That's where I start looking.
Three programs, up close.
Novartis
Transformation Lead on a €65M S/4HANA program in a regulated GxP environment. An 800-person SAFe setup across 7 business functions. I ran it top to bottom at once, from the program playbook down to backlog execution with two data teams, so the boardroom view and what shipped never split into two stories.
Nivea
Project portfolio manager on €45M of global SAP and data harmonization across 40+ countries. One rollout playbook, applied the same way each time. Country launches dropped from 12 weeks to 4, and data completeness went from 30% to 78%.
EDEKA
Program manager on EDEKA's AI marketing platform, currently steering it from PoC toward governed production, scaling to a roughly 25-person pod across marketing, tech, and AI. The hard part was never the model. It's brand-compliant content inside real guardrails, with quality gates and one owner per outcome.
Control, Flow, Production. One named program for each layer. See the full track record →
- 25+ engagements across 16+ clients in 10+ countries, on programs from €6.5M to €65M
- Ten years on enterprise transformations: SAP S/4HANA, ServiceNow, SAFe, regulated environments
- Doctorate (DBA): dissertation on organizational culture, applied as a working diagnostic toolset
- Currently completing Claude and Python certifications; building and running my own LLM workflows daily
- Available internationally, working US Pacific, US Eastern, and Central European hours
Three steps.
Intake call
30 minutes. Six questions. Written read-out within 48 hours. Yours to keep.
Scope note
One page. Recommended Phase 1 scope, price, duration, assumptions, risks.
Go / no-go
Within one week. Either we sign Phase 1 or we don't.
Happy to compare notes.
Whether or not we end up working together, the intake call is useful. Six questions. 48-hour read-out. Yours to keep.
Book the intake call →