Data Scientist Spotlight: Evan Lutins

Meet Evan. He’s one of our Data Scientists and an honorary developer here at Pinpoint. He’s currently a member of Team NaN, working on our issue and work forecasting to help engineering teams understand how long an issue or group of issues will take to complete. You can learn more about those here.

Evan is also a bitmoji enthusiast, and when we were still going into the office, he could be seen wearing a tie on self-proclaimed “Tie Tuesdays” — in which he was the only one to participate.

BREAKING UPDATE: Thanks to the pandemic and working from home, Tie Tuesdays are now “Tie-dye Tuesdays” in which he is still the only one to participate, and every day is Tuesday.

Click on the tabs in Evan’s Pinpoint Developer Profile above to learn more about his work. Get your own here!

So, tell us your story, Evan.

How and why did you end up in your role?

I was looking for a career change, did some research on the field of data science. I was interested enough to explore more and took an intensive General Assembly course, a 5-month program, 9-5 every day. Considering I had zero coding background and a brief statistical background, it did require a lot of self-teaching. Before I got to Pinpoint, I worked as a data scientist for car2go in the urban mobility space for about 1.5 years.

Describe yourself as a developer?

I am definitely still learning a lot about the engineering industry as a whole. I’m just a sponge over here trying to listen and take in as much as I can. The change to full-stack teams has definitely helped with the learning curve as I interact more with other developers with different expertise.

Tell us about your first dev job?

Honestly, this is probably my first “dev” job. I was a Jr. Data Scientist at my last company and was still analyzing data/building models, but the work was not nearly as engineering heavy.

What are you working on right now that has you most excited?

I’m working on a network graph model for Pulse, which is a personalized activity feed and using Natural Language Processing (NLP) to make recommendations across the product. We already have a lot of data from user's issue tracking systems and code repos (i.e., a specific repo, creator for a PR or project, who an issue is assigned to, etc.) we can use NLP to link similar PRs to Issues and vice versa.

These models have much broader use cases outside of Pinpoint, and applying it to our data has been really cool.

What is the hardest part of your job?

Understanding how everything in our internal engineering ecosystem from our pipeline to the data manager operates and works together. The more I understand our internal tools, the more confident I can deploy/roll out a new ML model. If something in regards to an ML model is breaking or throwing an error, I can more easily understand why and better debug. It also helps us better understand our possibilities and limitations as data scientists.

What does your team do that you think is unique?

Our Data Science guild has started a “Code Rodeo,” where we share all of the coding tricks we learn from our individual teams each week. One that I recently discovered was a VSCode extension, “Python Docstring Generator,” that automatically generates docstrings for python functions.

Auto Generate Docstrings

What do you prefer? Work from home, in the office, or some hybrid?

I’ve actually started working from my apartment buildings lounge room. I’ve found it helps drastically to just get out of my apartment. Since we have been remote, I’ve obviously cut down on commuting time, which is a great perk. It’s also nice to get outside and take my dog for a walk so I can break up the day and take a mental break.

What is your favorite project you have ever worked on?

I’ve enjoyed working on the network graphs mentioned earlier because they have much broader use cases outside of Pinpoint, especially in social media like Instagram and Facebook. It’s been challenging and interesting to apply a network graph to Pinpoint’s data. It makes you think about engineering as a social circle; developers, teams, communities, repos, and projects interact together in some capacity. The really cool part links similar developers based on projects, repos, and other developers they have interacted with. We can draw a lot of insight from this, and we will start to use the output from this model elsewhere in the project.

3 things you are loving? This could be a dev tool, podcast, a tv show, anything!

I just subscribed to a new emerging tech newsletter, which has introduced me to a lot of new tech.
I’ve also been spending a lot of time collaborating with the team on reimagining the daily standup. Which our first iteration of “The Daily” was released a couple of weeks ago.
I just discovered Dash, a python based tool for creating web apps. It’s a pretty flexible library allowing for a wide range of visualizations/dashboards.

Bonus! On a personal note, I have started playing Spikeball with some friends at Zilker park, and I think it might be the greatest game of all time.

Pinpoint Engineering

The TL;DR from my session on AI and EngOps at AIDe...