Hi, I'm Andrew. I'm currently building Void. I've previously thought a lot about algorithms, parallel computing, and quantum computing.
San Fran, CA
Building Void.
San Fran, CA
Built Glass.
New York, NY
I left APL to co-found DeriveIt. Our mission was to let technical people learn anything quickly: quantum computing, crypto protocols, ML architectures, etc. We ultimately built a site that prepares students for interviews at FAANG companies much faster, used in classes at top universities and praised by dozens of users.
I wanted this to evolve this into Comm2, but AI is already doing that.
Laurel, MD
At Johns Hopkins APL I led a project to estimate the resources needed to run a quantum computer. I also worked on statistics research, and my math on Beta distributions was published in a paper.
Remote
At Columbia I created a Mathematica tool to simulate & visualize wavefunctions and their energy bands near defects.
Ithaca, NY
I studied Computer Science at Cornell, with minors in physics and math.
I worked on parallel computing with Adrian Sampson (CUDA, Hammerblade), optimizing DSMM for the HammerBlade architecture. Also worked with Peter McMahon on quantum embeddings in ML.
Projects
reelers.io is a multiplayer io game I built on top of an html canvas & websockets. It can take a minute or two to boot, but it's pretty fun, especially with friends. Here's a clip of it on GitHub.
I wanted to write a 3D renderer without using any graphics pipelines, so I came up with some math and wrote 3DTest.- custom-markdown parses a text file and renders the result in React. I wrote it as a simpler alternative to MDX and TipTap so that I could fully customize Markdown's syntax.
While working on DeriveIt, I inspired my brother to build Recursion Playground, where you can visualize recursive functions as self-similar fractals.
Misc projects. At some point I made a blocky physics game in Java, and an escape-the-room game in flash. I also animated a few things for fun.Life goal: I think you should be able to prove P != NP using a new kind of information theory.
You probably know that we can use information to speed up an algorithm's time complexity; like if you have the information that a list is sorted, that often lets you solve a problem faster.
I'd like to go the other way around, and answer questions like how much "information" did we gain by knowing the list was sorted? You can use complexity to figure this out.
If you can come up with a relationship between complexity and information, and show that NP-Hard problems are asking for exponential information, then you'd know their complexity is exponential too.
I have some early results for certain types of polynomial time problems and would love to find some time to write about this topic.