Second Brain: Inspired by tools like Pinboard and the concept of Vannevar Bush’s Memex, I built a tool to intelligently store, search, and interact with personal documents.

Powered by Vespa + Mixtral-8x7b, this product autonomously did some of my school and personal grunt work and performed better than ChatGPT for Q/A over documents.

Penn Clubs: Penn Clubs is the official student club repository at Penn. While in school, I worked as a developer lead on the product, which is managed as part of the umbrella student club Penn Labs.

Check us out! We build a lot of cool, open-source software for the Penn community and have saved the university well into the 6 figures in software costs.

Spruce: A friend and I wrote a parser and interpreter in Haskell for a language called Spruce with functional features (first class functions, lexically scoped closures) and low-level concurrency primitives (fork, wait) alongside a simple interface to support transactions over memory.

This project makes it easy to write (in Spruce):

let shared x : int = 0
func f() -> void {
   for (i : int in range(3000)) {
      atomic {
          x = x + 1;
      }
   }
}
s1 = fork(f); s2 = fork(f);
wait(s1); wait(s2);
assert x == 6000

and guarantee consistency in output. The STM monad is beautiful.

Distributed ML Pipeline: For a DevOps final project, I built a distributed Machine Learning pipeline (training and inference) on Kubernetes that aims to be model-agnostic and supports MLOps best practices (Continuous Training and Continuous Monitoring).

CIS 1880: DevOps: I co-taught a (now discontinued) class on DevOps at Penn. Born out of an older NETS major’s distaste for Penn’s relatively outdated cloud computing class – NETS 212 – this class cultivated my budding interest in cloud infrastructure. Unikernels anyone?

RuneScape 2D: I used to play a lot of RuneScape when I was in high-school, so for a class final project I built a minimal 2D version of the game’s old tutorial, mostly for nostalgia.