I write algorithms that tune large-language models.

I developed optimization algorithms and machine learning techniques that help people make optimal decisions.

Celerie is a simple, privacy-focused calorie counter. I am the sole developer. I wrote it with the following technologies:

• SwiftUI was used to write the user interface.

• Itegration with Edamam provides the food database. (Users can create custom foods for missing items.)

• Combine processes network requests and user events.

• CloudKit syncs data between a user’s devices.

I wrote software that many of the world’s largest companies (Boeing, Panasonic, BASF, and others) use to accelerate their chemicals and materials development. At Citrine, I led efforts to develop global optimization and machine learning algorithms specifically for materials. Some problems I’ve worked on include:

• co-inventing a patented method to featurize materials,

• training random forests with little data (each data point can cost >$1M and take several months to acquire),

• developing optimization algorithms to find materials that meet target objectives(I implemented an algorithm that improved the quality of suggested experiments by 100x), and

• quantifying uncertainty (experiments are expensive, I helped develop methods that suggest experiments low uncertainty or high potential).