Hi! I'm Roujia Wen

I am a master's student in applied mathematics at the University of Cambridge.
Previously I was a member of the founding class at Minerva Schools studying physics and computer science.
I'm interested in applying quantitative tools (optimization, machine learning and inference, mathematical and computational modeling, algorithms) to challenging problems at the intersection of technology and social good.

academic research

Self-organizing Behavior

self-assembly of active Brownian particles (2015, 2018-2019)

I wrote my undergraduate thesis on the self-assembly of active Brownian particles. I developed agent-based simulation software equipped with order parameter measurements and an interactive genetic algorithm, enabling the efficient exploration of parameter space and the inverse design of systems with desired emergent behaviors (supervised by Prof. Eric Bonabeau and Prof. Rohan Shekhar).

paper   summary article   source code   web app (lite)
Quantum Information

numerical optimization in quantum information (summer 2018)

I participated in the undergraduate research program at the Perimeter Institute for Theoretical Physics and worked with Dr. Denis Rosset on numerical formulations associated with the inflation technique—an algorithmic procedure to derive Bell-type inequalities for quantum networks.

talk video   slides
Muon Identification

deep learning in particle physics (fall semester 2017)

During a semester in Korea, I worked with Prof. Tae Jeong Kim at Hanyang University Particle Physics and used deep neural networks to improve the identification of isolated muons from simulated collision data.

The SAT Problem

algorithm for NP-complete problem (summer 2016)

As an REU at the Santa Fe Institute, I developed and optimized a local search algorithm for the boolean satisfiability problem (SAT) by combining the novelty search heuristics and multi-objective optimization (supervised by Prof. Joshua Grochow).

talk video   slides   paper

industry projects

VAD Visual

scientific computing & visualizations (summer 2017)

I programmed software that post-processes data from large scale biofluid dynamics simulations, calculates statistics and provides interactive visualizations.

screenshots  poster
Cruise App

machine learning & recommender systems (gap year 2015)

I gathered tens of thousands of customer reviews, extracted feature-based sentiment scores using machine learning and designed a recommendation algorithm for cruise matching.

web app




For our first-year final project, I worked with three other Minerva students on a project proposal to combine active learning and interactive exploration tools with Wikipedia content. We presented at the Wikimedia Headquarter in San Francisco.

talk video  slides