About
Download ResumeMy path to healthcare decision science ran through mathematics, psychology, and academic research on how people decide under uncertainty.
I studied mathematics at a liberal arts college in Iowa and earned a PhD in psychology from a research university in Switzerland, where my research focused on judgment and decision making. After postdoctoral positions at research universities in Germany and Switzerland, I moved into industry as a senior data scientist at a global pharmaceutical company, then as a senior quantitative scientist at a healthtech company using real-world data to improve cancer care. In 2022 I co-founded a healthcare analytics consultancy, where I served as CTO before taking a staff quantitative scientist role at the largest psychiatry practice in the US. I now run an independent consulting practice focused on decision support, data infrastructure, and technical training.
One idea runs through my work: the best way to make decisions under uncertainty is with simple tools approved by experts.
This is true for decisions ranging from clinical decisions in the emergency room, to strategic decisions in the board room. In both settings, the people making the call have to trust the logic, recognize when a tool does not apply, and adjust as evidence evolves. That view was shaped by my research on fast-and-frugal decision making and led me to build open-source software for constructing algorithms in fast-paced, high stakes decisions. It has been cited over 100 times and applied to problems including radiation dose triage for lung cancer, lung cancer screening criteria, and clinical course prediction in respiratory disease. I bring the same philosophy to every engagement: reduce a problem to the signals that actually drive action, and make the reasoning inspectable by the people who have to live with the decision.
That principle shows up in the applied systems I have built for pharma, health tech, and startup clients.
At a cancer-focused healthtech company, I designed and led development of an Oncology Analytics 'Kitchen', an internal package that replaced a fragile, undocumented data pipeline for oncology real-world evidence. Standard analyses that used to require more than 100 lines of code now ran in fewer than 10, and the system became the team's analytical backbone. I presented the work at rstudio::conf 2020. More recently, I led MVP design and development for CHORD, a health-economics decision engine built on longitudinal claims data. Working alongside engineers, epidemiologists, biostatisticians, and senior leadership, I helped cut the average client time to a go/no-go decision from 7 days to 30 minutes. Across projects like these, I move between strategy, design, and technology, translating a client's decision problem into a system that supports it and adapting both as requirements evolve.
AI has the potential to improve every stage of decision making, from identifying problems to designing and implementing solutions.
I am optimistic about what AI can accelerate in the kind of work I do, from cohort definition and study design to the translation of scientific questions into production systems. That potential is real, but it does not change the fundamentals. An AI tool has to earn its place the same way any other tool does, through rigorous testing, validation against known ground truth, and ongoing human oversight. This matters especially in healthcare, where the cost of a quiet error can be high. I help clients figure out where AI can genuinely add value, where a simpler model is the safer answer, and how to build the monitoring needed to tell the two apart over time.
Teaching and communication shape how I deliver technical work.
I have trained more than 500 students and professionals in statistics, data science, and decision making, including through a bootcamp I co-led for professionals in pharma and finance. I have given invited talks at rstudio::conf, useR!, and R/Pharma, and I have a peer-reviewed publication record in decision science and health outcomes. That teaching habit carries into how I engage with clients. I explain my reasoning as I go, document my work as I do it, and push back when I think a question is being answered the wrong way. Getting the answer right matters more to me than looking right, so I would rather flag a problem early than deliver a polished result that misses the point.