Alex Bertsch


I'm Alex.

I like to make cool things and play with data.

I'm currently finishing up an ScB in computer science at Brown University.

I've worked at companies like Yahoo and Dropbox, designing and building a wide variety of projects. Some of the most interesting things I've done can be found further down this page.

This site shows off some of the more impressive things I've made. It may also contain a blog depending on whether I'm actually inspired enough to actually write anything.

My work

Brown University – Teaching Assistant

CS100: “Data Science for non-Concentrators”.

I'm currently helping teach a course designed to introduce the power of data science to non-computer science majors.

Fall 2016

Dropbox – Software Engineering Intern

Designed, implemented, and open sourced a distributed security alerting system.

To increase the effectiveness of incident detection, I designed a distributed alerting chat bot that reduces the workload of the detection and response team. The bot reached out to users directly, verifying that certain actions were done intentionally and not as the result of a security breach.

June–August 2016

Brown University – Teaching Assistant

CS1951A: “Introduction to Data Science”.

Designed assignments, led review sessions and ran labs for a course designed to showcase the various aspects of data science. Primarily taught in Python, with a minor focus in SQL and D3.js. Full course website.

Spring 2016

Yahoo – Software Engineering Intern

Designed a new instrumentation library for use with Yahoo's monitoring system.

Designed a library for various metrics to be used with Yahoo's internal monitoring system. The library was created to be paired with Yahoo's monitoring system refresh.

June–August 2015

My projects


June – August 2016

Extensible distributed alerting chat bot.

Securitybot is an open source and extensible realization of a simple method to increase incident detection effectiveness. Built on top of generic messaging, 2FA, and alert management plugins, Securitybot connects to any existing infrastructure to automate to process of investigating incidents. Rather than wastes security analysts' time on false positives, the bot reaches out to users for them, quickly asking whether or not they performed an action.

Atari Vision

January – May 2016

Vision-based reinforcement learning on Atari games.

With inspiration from Deepmind's similar paper, I helped create a vision-based approach to learning how to play Atari games, namely Space Invaders. Initial implementation resulted in an agent that could effectively best random actions or event a simple naïve agent. Work was later adapted to use a neural network to achieve better results. See the source.