Dreaming in code with Atilio Barreda
An interview with Atilio Barreda, Processing Fellow 2022.
Could you tell us a little about yourself?
I’m Atilio Barreda, a Peruvian software engineer and adjunct instructor at the City University of New York (CUNY), New York City College of Technology in downtown Brooklyn.
What was your fellowship project and how did it begin?
My fellowship project was the beginning steps of a whiteboarding app that is focused on data science and machine learning concepts.I got the idea, because I started teaching remotely at the beginning of the pandemic. There are actually lots of great things about remote learning I found, but illustrating concepts visually was not one of them. So we do have some whiteboard apps available to us, but in some cases I felt like I had to be like a spectacular digital artist, which I wasn’t, to showcase relatively simple concepts. I just got in my mind that I wish there was a tool that was pre-built for teaching data science concepts and generating data visualizations based on that.
After some research, I realized that such a tool didn’t exist. And then I thought, wait, I’m a web developer, so maybe I could do it. So that’s pretty much how it all started. I was hoping to do it anyway, but then when I found the opportunity to get support, I was really hoping that I would get it. It was just fantastic that I was able to devote time to making it happen.
The whiteboard is one big kind of Processing context instance, and then I used ml5js to do the data analysis and the number crunching. What I did is I essentially fed back whatever happened in ml5js into a way that p5 could understand and generate visualizations from.
It is currently in a stage where there is one data science concept that I’m illustrating visually, a type of clustering called k-means clustering. I have a public demo site, where you can go select a specific clustering method and then generate visualizations based on randomly generated data. That’s where the project is at now. My goals are to integrate a few more methods and include, in addition to randomly generated data, some sample data sets with real life applications to maybe make things more understandable.
Do you have any advice for future fellows, including how to structure your time?
I feel like I was in a pretty lucky spot in terms of timing because I was able to not take on additional freeland work and just replaced it with time to work on the project. I definitely had mini sprints during the week. I think I would spend around seven to like twelve hours a week and it would usually be two or three days in a row because that’s how I find I become most productive. I’ll work forever and then I’ll force myself to go to sleep and then I’ll dream about code and then I’ll do that.
Do you ever find the solution to something you’re stuck on in your dream?
It’s so strange, I’ve definitely had “Eureka!” moments while daydreaming. I guess I should not code to code more often I guess. My best advice for coding is not to code.
Could you tell us a bit more about who your students are, and what’s the context that you’re teaching data science in?
I teach at a public institution and we are a Hispanic serving institution. I feel like so many of the students don’t have the capacity to devote their full time to study. And I don’t want to badmouth New York or the infrastructure because CUNY is doing amazing things, but we really don’t have access to a lot of what the private institutions have access to. I am building this tool for other instructors, and thinking about ways to make educational tools that are actually accessible and open source on Github. For example, other instructors could use this tool and modify it for their classroom needs.
A lot of data analysis tools are proprietary and as a public institution we can’t really give the students the access we, as instructors, want to. I’m a small part of this open source educational tool community, but it’s really important to invest in, especially if you can’t compete with everything that big private schools have.
For example, some professors might use only Excel. Excel files are widely used in the data community, but it’s still a proprietary format. It’s basically a proprietary version of a CSV, comma separate values file, but why not just send a CSV? It’s important to err on the side of the most accessible file format.
Do you have any words of advice for future fellows?
I think Processing does a good job of setting expectations. But no matter what, whenever you get something like a fellowship there is a moment when you think “This person is my boss!” But that is so far from what the truth is. My mentor, Abram, was really supportive and everyone was just there to help me. Don’t be afraid of saying you don’t know what to do next to your mentor — my mentor was not only well versed in using p5.js but was also an educator.
Atilio Barreda II is software engineer, adjunct lecturer at CUNY City Tech, and MS student in Data Visualization. http://cv.atil.io. Follow Atilio on twitter.
