Sixteen years inside manufacturing systems.

My first job was industrial engineering at an electronics plant in India in 2010. Time studies. MOST analysis. Line balancing. Watching an operator repeat the same reach-and-place motion 400 times a shift and asking why the bin was 18 inches too far away.

That work isn't glamorous. It is also the reason I can walk onto a site today and know within an hour whether the daily management board is real or theatre.

At Jabil I moved from industrial engineering into Lean Six Sigma leadership across contract manufacturing operations in India, Malaysia and Singapore: enterprise storage and networking programs, build-to-order and configure-to-order cells, rack assembly, automated systems. Over six years I ran 54+ Kaizen events and 32 major Lean Six Sigma initiatives. We converted 15 cells from batch to single-piece flow and cut WIP by 92%. OEE moved from 32% to 56%. Throughput improved around 60%. The work generated over USD 900K in annual savings.

The number I'm proudest of isn't any of those. It's 300+, the people we developed through training and coaching, 50+ of whom certified formally. Projects end. Practitioners keep going.

I joined Thermo Fisher Scientific's BioProduction division during a rapid manufacturing scale-up. Site productivity was around 40%. Over roughly eight months we took it past 90% through the PPI Business System, daily management, value stream work and a lot of unglamorous constraint removal. OTIF improved about 40%. Inventory came down about 60%. Warehouse and 3PL redesign took out roughly USD 95K a month.

Here is what I learned there: a scale-up doesn't fail on strategy. It fails on the fourteenth handoff nobody mapped.

At Roche External Manufacturing, I lead Business Excellence across a global network of 180+ contract manufacturing partners. Roche is the sponsor. The partners are separate companies, with their own priorities, their own leadership and their own definition of urgent. I have no authority over any of them.

So the tools change. You govern instead of direct. You build scorecards people actually believe, operating reviews that produce decisions rather than status, and escalation paths that work when the person who has to act doesn't work for you. I've built and govern a CHF 52M+ cost-of-inefficiency opportunity pipeline across External Manufacturing and delivered CHF 400K+ in hard savings from workflow standardization, redesign and automation.

I keep those two numbers strictly apart. A pipeline is a promise. Delivered savings are a fact.

I've written two books about it. I'm doing a part-time PhD at the University of Waterloo on data-driven operational excellence in regulated manufacturing. I build and test AI tools for investigation and root cause work.

And my position hasn't moved in ten years: optimize before you automate. Point a model at an undefined process and you get expensive, confident nonsense. Your process is the prompt. Fix the process first.