How a geochemistry lab at Notre Dame became the origin of everything I do in infrastructure today.

The question I get most often when people learn about my background is some version of: how did you end up here?

The résumé looks like a series of hard left turns; geology, then network consulting, then a startup, then a decade at IBM, then SaaS infrastructure. The only non-developer on a software architecture team. It doesn’t look like a straight line.

But it is one. It just requires the right lens to see it.

In 1994 I started a master’s program in inorganic geochemistry at the University of Notre Dame. My thesis work involved developing an analytical procedure for quantifying platinum group elements (platinum, palladium, iridium, rhodium, ruthenium) and gold in geological samples.

The field had a problem; these elements are present in rocks at extraordinarily low concentrations. Femtograms per gram, or parts per quadrillion. The conventional answer was better instruments. More sensitive machines, lower detection limits, more sophisticated signal acquisition. The whole field was oriented toward the instrument as the solution.

I looked at the same problem and saw it differently.

The instrument wasn’t the constraint. The matrix was. A dissolved rock contains so many other elements at so many higher concentrations that they mask the signal you are trying to measure. The platinum was there. It was simply invisible behind everything else.

The detection challenge was not measurement. It was separation.

Remove the matrix first. Use cation exchange pretreatment, a resin that binds specific elements and lets everything else pass through. This allowed us to deliver a cleaned, concentrated fraction to the instrument. Now a less exotic machine did the job cleanly, because the hard work was done upstream. Simpler. Cheaper. Reproducible.

This procedure worked. It worked well enough that other researchers adopted it. My name shows up in the acknowledgments of papers I didn’t write, which is the quiet version of success in academic science.

I did not fully understand at the time that I was building a methodology, not just a procedure. The technique was specific to geochemistry. The problem-solving was not. Change the lens, and what was invisible comes into focus.

The work was not just analytical. Running the ICP-MS meant overnight sessions alone in the lab. Plasma temperatures required continuous cooling, and skimmer cones would burn through if the chiller failed. The samples, in some cases, were lunar regolith. Irreplaceable.

I slept on the lab floor so I could hear if the chiller failed.

When the sample cannot be replaced and the instrument cannot fail, you find every point of vulnerability before it becomes a problem. You do not wait for the alert. You make sure the alert never fires. I have been doing the same thing in infrastructure ever since.

The most important finding from the thesis work was not the procedure. I was looking at the wrong layer.

My replicate analyses were showing more variability than expected. The obvious assumption was instrument error. It wasn’t. It was sample preparation. Platinum group elements concentrate in microscopic grains scattered unevenly through the matrix. The instrument was reading correctly. It was reading different things.

The problem was upstream of where I was looking.

That pattern has followed me through every role since. The constraint is never where the field assumes it is. Change the layer you are interrogating, and the answer becomes visible.

I left the lab in 1996. The methodology did not.

I spent a decade at IBM building the workload analytics program for one of the world’s highest-scale MDM platforms with our largest customers processing billions of records, running trillions of probabilistic comparisons. The performance challenge was how do you find a signal buried in a noisy system where the thing you are looking for is a small fraction of the total?

You characterize the matrix. You isolate the signal. You remove what is masking it and let the instrument do its job.

The visualization techniques I built could plot every customer transaction in time series. The standard practice of plotting averages was hiding the real variability behind an aggregate that looked clean. If you can see every data point, you can see the outliers. If you can only see the mean, they disappear.

Same lens. Different matrix.

My current role extends the methodology further. I build models that take controlled lab measurements and translate them into multi-year cost projections for a finance audience. I quantify new product workload impacts projected from test results. I drive infrastructure investment decisions based on measured data rather than vendor estimates.

The matrix is now a global SaaS fleet. The signal is cost, performance, and risk. The instrument is a model that has to hold up to scrutiny from engineers, finance teams, and executive leadership simultaneously.

Same lens. Different scale.

This is not a career story. It is an explanation of a methodology that originated in a master’s thesis about platinum group elements. The contexts have changed completely. The underlying structure of the work has not.

There is always a matrix. There is always a signal. The skill is knowing where to look.