About Me
I'm a mathematician, computer scientist, and an engineer.
My background is in linear algebra and mathematical optimization. I love optimization because most business and engineering problems are easily expressed as optimization models, and these models can be solved with powerful algorithms.
I currently lead the cuOpt solver team at NVIDIA, working on GPU-accelerated optimization. Before that, I spent five years at Apple, first building real-time optimization solvers for autonomous systems and then as an engineering manager leading a trajectory optimization and ML prediction team. Earlier, at Gurobi Optimization, I worked on barrier and crossover, MIP heuristics, and developed the automatic tuning tool and the Gurobi Instant Cloud.
Selected Writing
- Batched First-Order Methods for Parallel LP Solving in MIP Honorable mention, 2026 Harvey J. Greenberg Research Award, INFORMS Computing Society
- Solve Linear Programs Using the GPU-Accelerated Barrier Method in NVIDIA cuOpt
Projects
Below are some projects I've worked on. Many of them are fun, some are a bit more serious, all are technical.
Routing planes around thunderstorms: During my postdoc I implemented an optimization model of the National Airspace System that included: nearly 4000 flights, all high-altitude traffic sectors, and the 34 largest airports.
I used the model to analyze a day when many flights exprienced delays due to weather and air traffic.
Logo Interpreter: An interpreter for the Logo language. The interpreter is written in OCaml and compiled to Javascript, so you can run Logo in your browser.
Colorization on GPUs: I implemented a method to color a grayscale image from a small set of color annotations. The method is based on a parallel implementation of preconditioned conjugate gradient in CUDA.
I used this method to solve two linear systems, each with 3 million unknowns, on two GPUs. The solutions produced a coloring of a 1050 x 3360 image of El Capitan and the lower Yosemite valley.
Low-polygon 3D model of my head: For Halloween in 2011, Jim Paris and I wrote some custom software to create a low-polygon model of our heads from a series of still photos.
We printed and assembled the models and used them for Halloween masks.
Brachistochrone: I used optimization to compute solutions to the famous brachistochrone problem.
I used Chebyshev spectral collocation to approximate the optimal-control problem. The resulting nonlinear optimization problem has linear-inequality constraints. To solve it, I developed my own quasi-newton solver with a primal-dual interior-point method for the QP subproblems.
Optimal Bidding in Power Markets: As a research assistant at EPRI, I analyzed the effects of optimal bidding in a simple power market when a vertically-integrated firm (controlling generation, demand, and electrical transmission) seeks to maximize profits.
Heliochronometer: A sundial that tells standard mean time by computing a correction from local real time via a mechanical computer.
The heliochronometer, which I built for Stanford's Design and Manufacturing class, is based on the design of a 1906 brass Pilkington and Gibbs heliochronometer.
Interactive Oregon Trail Diary: In April of 1852, my ancestor, Philemon Morris, set out on the Oregon Trail. He kept a detailed account of his trip west, and his first two winters in Oregon, in a small diary.
I wrote a small web app that matches each entry in the diary to my best guess at his location along the trail.
Astrophotography: I recently started taking pictures with a Seestar S50 Pro telescope on my roof. It's amazing to see nebulas and galaxies from my light-polluted house.
Herreshoff/Doughdish 12.5: When I was 12, my grandfather and I tried to build a wooden Haven 12.5 sailboat. We didn't get very far. We steamed the oak ribs, bent them over the plywood frames, and came back the next day to find them all broken. So this sailboat has long been a dream of mine.
In 2026 I was delighted to buy a Doughdish, a replica of the Herreshoff 12.5 with a fiberglass hull. Our family had a blast sailing around Wellfleet and Beverly Harbor in this small, beautiful sailboat.
Droning: I'm a big drone enthusiast. I made a few videos you might enjoy watching:
- In Massachusetts: Hingham and Hull through the eyes of an Inspire 1
- Castle Hill by Drone - The Crane estate in Ipswich, MA
- Flying the Phantom around the North Shore
Lego Orrery: An Orrery is a mechanical model of the solar system. My orrery depicts the motion of the Earth-Moon-Sun system. Four independent rotations are powered by the rotation of a single axel powered by a Lego 71427 motor.














