Hi there, I'm Allen! I help companies collect their environmental data and manage their resources at Nectar.
Previously, I studied at MIT where I helped organize HackMIT as Dev Head, where I managed teams and
shipped open source tools. I was at some point the Co-President of MIT Poker Club.
I love meeting folks from all walks of life. Come say hi 👋 and email me!
What are you up to now?
I spend most of my days building Nectar where I write software, help large companies with their environmental resources, and meet brilliant minds who are dedicating their lives to solving climate change.
What do you do for fun?
A mélange of things... I enjoy reading about science and how people
I get excited about making useful things for people and creating cultured communities. I like
tipsy teammates with my mediocre card magic, making music with friends, and occasionally writing.
While the vast majority of my time is of course dedicated towards Nectar, in the little time I have outside, I've been helping maintain Dialogic as an on-and-off project with Antonio Frigo, Jonathan Yin, and Joshua Mishrikey. If you're interested in what I'm working on, or would like to join the fight towards rapid decarbonization, feel free to reach out!
Dialogic: Tracking Seminar-Style Discussions
I helped design and build software to track what happens in a group discussion in a classroom.
The project was done in conjunction with Antonio Frigo, Jonathan Yin, and Joshua Mishrikey.
Over COVID, the math competition I had started in high school ran into issues. Antonio Frigo and I teammed up to build a platform that allowed folks from my high school to run a contest online complete with automatic grading and a leaderboard.
I designed and built out the first version of ballot, HackMIT's virtual judging system, with a wonderful dev team. The app uses graph theory and probabilistic algorithms to optimize scheduling and infer rankings.
I implemented an MCMC-based cryptogram solver for MIT's inference and information course. My report was recognized for its unique approach of using a symmetrized form of KL divergence to find a better initial state.
After the course, I exported half the project into an app. Now it runs outside the CLI 😀.
We implemented cache-oblivious indexing for matrices using row-major, hilbert, and z indexing and benchmark cache misses using valgrind on an EC2. To help visualize the indexing schemes, we built a static site to "walk" along the matrix.
The research project was conducted in collaboration with Jett Wang and under the mentorship of Erik Demaine.