Home

Back in middle school, I wrote a Vigenère cipher program so my friends and I could pass secret messages, and I've been hooked on programming ever since. I went on to study Computer Science and Philosophy at the University of Chicago, and I've spent over six years working in software and cybersecurity: investigating breaches at Stroz Friedberg, building software for clients at Manifold, and now at Meta, where I've worked on Instagram ads and AI model training.

I'm drawn to algorithmic problems and back-end systems, but my experience spans the full stack, from front-end development to data modeling and security.

Outside of work, my interests include running, soccer, building custom keyboards and PCs, music, and the outdoors. For more about what makes me tick, check out my interests page.

I also write occasionally about technology and whatever else I'm thinking about. You can find my posts on my blog.

Below, you’ll find summaries of my professional experience and side projects. If you'd like to get in touch, feel free to send me a message.

At Meta, I spent my first year and a half on Instagram's ads delivery team, improving how Instagram selects and caches the ads it considers for each user. Much of the job was experimental: designing live experiments and weighing competing metrics, like revenue and computing cost, to decide which changes to launch. I also built monitoring and alerts for our live systems and took part in the team's on-call rotation. Since June 2026, I have contributed to training Meta's AI models.

At Manifold, I spent most of my three and a half years on a single healthcare client engagement. I led a security overhaul that unified control over who could access which data across dozens of the client's services, built key parts of its web portal serving over 100,000 users, and regularly explained technical trade-offs to the client's project managers. Along the way, I worked across the stack: React and Angular front ends, Scala and Python services, relational and graph databases, and cloud infrastructure on AWS.

At Stroz Friedberg (then part of Aon Cyber Solutions), I investigated ransomware attacks, business email compromises, and intellectual property theft, reconstructing what happened from digital evidence and explaining my findings to engagement managers and clients. I also built Python tools to preserve GitHub repositories as evidence and to identify personal information exposed in breaches, which supported many client investigations.

Klotski Solver is a web app I developed that enables users to create, solve, and view optimal solutions for Klotski puzzles using an implementation of the parallel Breadth-first Search algorithm. The project originated as a C++ command-line tool and has evolved into a comprehensive React/Redux web application. The backend, developed in Rust with the Axum framework, interacts with a PostgreSQL database. Both the user interface and the API are containerized with Docker and deployed on Fly.io.

Check out my blog post Cracking Klotski for more information on the solver algorithm.

Front-end source code

Back-end source code

I created the portfolio website you are currently viewing using ElysiaJS on the Bun JavaScript runtime. I used HTMX and TailwindCSS to enhance site interactivity and styling within individual TSX elements. I employed Vercel Functions to fetch data from the Strava and Spotify APIs for use on the interests page. The site features a robust admin panel that enables the creation, deletion, and modification of home page content and blog posts, utilizing Turso SQLite for efficient data management. The user interface is containerized with Docker and deployed on Fly.io, while the serverless functions are hosted on Vercel.

Front-end source code

Back-end source code

Tvrtl is a betting simulation I developed with my friend Stefan Tesliuc where players place bets on the path of a turtle navigating from the center to the edges of an $n$ sided polygon. The turtle’s random trajectory is modeled using Brownian motion and smoothened using spline curves from Three.js. The app, developed in React Native, employs Recoil for state management and NativeBase for styling.

Source code

Info.rm is a Django web app I developed that displays a dynamic dashboard showcasing the seasonal performance of soccer players across Europe's top leagues. Additionally, users can craft customized views by applying a wide range of filters on players' in-game statistics and personal attributes. The web app pulls its data from API-Football and manages it using the SQLAlchemy ORM and SQLite.

Source code

ElmCalc is a web-based calculator I developed using Elm. The calculator's design is inspired by the iconic TI-84 and it performs basic arithmetic operations. I created this project for my college Functional Programming class, and it was selected to be featured in the class showcase.

Source code