Executive Director · Applied AI/ML · Agentic & Generative AI · NLP · PhD, Biophysics
I lead teams that build AI agents and generative AI tools people actually want to use. I started out as a scientist, and I still love the moment an idea turns into something that works.
I'm an Executive Director of Applied AI/ML at J.P. Morgan Chase in New York. I joined the Machine Learning Center of Excellence as its first hire, and it has been a joy to watch it grow into an AI organization of more than 200 people.
These days my teams build AI agents, generative AI, and language tools that take the busywork out of people's days. I spend a lot of time with the people who use what we build, because the best ideas usually come from watching how they really work. I've also learned how to show clearly what difference the AI made, which is one of the harder and more rewarding parts of the job.
Before AI, I was a scientist. I earned my PhD in biophysics at Johns Hopkins studying how proteins fold, and that curiosity has followed me ever since.
200+people in the AI organization I joined as its first hire
10+years building machine learning and AI
2granted U.S. patents, with more pending
PhDin biophysics from Johns Hopkins
02 / Focus
What I'm working on
/01 · featured
Agentic AI
AI agents that can plan, use tools, and know when to hand things back to a person. I care as much about how we test and watch over them as about how clever they are, because trust is what gets an agent used.
/02
Document generation with RAG
Retrieval-augmented generation that drafts well-structured documents from trusted sources. The patent has the details.
/03
AI in everyday workflows
Language models that help people spend less time on routine steps and more on the conversations that matter. See the patents on improving sales processes and AI-assisted pricing.
/04
Evaluation & feedback
Feedback loops, monitoring, and LLM-as-judge evaluation that keep models improving after launch, plus synthetic data for training them.
/05
Growing teams
Hiring, mentoring, and growing people across machine learning, engineering, and product. Watching someone on the team take on something they didn't think they could do is my favorite part of the job.
My PhD at Johns Hopkins asked how proteins fold, one repeat at a time. Along the way I grew protein crystals, solved the structure of a new kind of repeat protein, and built models to understand what holds it together.
PDB · 4Y6W
A new class of TPR-like repeat protein
The structure is a kinesin-light-chain-like protein from the fungus Podospora anserina. Its repeats are nearly identical in sequence, which makes it an unusually clean model system for asking how neighboring repeats stabilize one another.
Growing the crystals. Protein crystals under the microscope, before they were frozen and shot with X-rays. Getting a protein to pack this regularly is often the hardest step.From structure to energetics. Left: electron density resolving a Glu30–Tyr58 hydrogen bond. Right: chemical denaturation of constructs with different numbers of repeats. Adding repeats shifts unfolding to higher urea, the stabilizing effect a nearest-neighbor Ising model quantifies.
/model
Nearest-neighbor Ising analysis
Each repeat is treated as folded or unfolded, with its own stability plus a coupling energy to its neighbors. Fitting many constructs globally with shared parameters separates the two terms.
/software
Open analysis tools
The fitting programs, with bootstrapped confidence intervals, were published for the community in Protein Science (2021).
/methods
Biophysics toolkit
Crystallography, analytical ultracentrifugation, circular dichroism, fluorescence kinetics, and Monte Carlo error analysis.
I grew up on story time and board games, and both stuck.
Games & strategy
I love board games and strategy games of all kinds, and I've always been curious about the math behind them: what each move tells you, and how to find the best one.
Where it started Story time with my mom.
/languages
Dutch & French
I enjoy learning other languages. Right now I'm working on Dutch and French.
Ready to roll/on the move
Rollerblading, running & volleyball
I love rollerblading and running, and I play beach volleyball whenever the weather allows.
/play
Games
A few games you can play in the browser, with notes on the math behind them. Mastermind is up first.
