A Day in the Life of a Machine Learning Engineer Intern

When I stop and think about it, life is a bit crazy at the moment. I’m standing at the threshold between the end of my academic life and the beginning of my professional life. I’ve just graduated from college, moved into a house with some friends who have an impressive recording studio in their basement, and started my first ever internship with The General® as a Machine Learning Engineer (MLE) Intern. I am absolutely thrilled to have been chosen for this internship, as it is the first internship that the Data Science team has ever offered! But what exactly does a Machine Learning Engineer Intern do? Let’s talk about my typical day.

I try to wake up at 6:00a, but it’s more likely to be anywhere between then and 6:30. I like to leave for work at a consistent time, so breakfast is either a bowl of cereal or eggs and turkey bacon, depending on the number of times I hit the snooze button. My commute to work isn’t bad at all, around 30 minutes to travel the 16 miles along Donelson Pike. Fortunately, my route avoids the typical bumper to bumper Nashville traffic and treats me with a scenic view of the airport’s runways and rolling green fields. On occasion, I even get to watch a plane land (which I must admit is awesome to see).

Once at work, I have a typical routine I like to prepare myself for the day with. The interns are grouped together on a different floor than the Data Science team; so I’ll generally clock-in, put my lunch in the fridge, and check-in with the team. I’ll grab coffee on the way back to my desk and begin my day. Here’s what my setup currently looks like:


 
 

I was provided with two monitors, and I like to leave my laptop open to have a triple screen setup. The laptop is a smaller screen and out of my direct line of sight; reserved exclusively for emails. I use the two main monitors for a variety of tasks. Typically, I’ll put chrome on the left, where I have quick access to my Trello board or any information I need to look up. The right screen usually has PyCharm or a Jupyter Notebook, depending on what I am working on. Back to my morning preparation, I first look at Outlook to see if I have any new emails and check my calendar for the day. Then, I check my Trello board:


 
 

I have used Trello before while developing software projects, but Tim Dobbins actually showed me how he uses it for all aspects of his work. I find it helpful for keeping track of everything that I have going on and that it allows me to reach mini checkpoints to self-validate that I am being productive and prioritizing correctly.

Once per week, the Data Science team has a meeting where all the Data Scientists (DSs) and MLEs come together and discuss a variety of topics. As the manager, Chris will orchestrate the flow of the discussion, keeping things organized and efficient. First, he addresses any new changes to the company that impacts our work. For example, a change to the company’s data pipeline or infrastructure could be a topic that is investigated at this point. We are usually notified of these topics through email in advance. Then, we discuss current projects. These check-ins keep everyone on track and facilitate the communication and project handoff between the DSs and MLEs. The two roles are intertwined from idea conception to production, so it is essential that the entire team meets to address issues and track progress. Finally, time is left at the end of the meeting to address issues, concerns, and any other topics that were not planned out. The meeting then breaks and everyone returns to their work.

This summer, I will be working on a fascinating project researching and developing methods for model monitoring and anomaly detection. At the simplest level, all models are built off of data and make predictions based off of the data they learned from. By creating these methods, we will be able to see if a model is not predicting as well as it has in the past. This information will be used to determine a retraining schedule and compare different model versions. In the next phase of my research later this summer, I will build a pipeline that automatically adds data to this monitoring service; taking appropriate actions if issues are detected and notifying team members accordingly.

Since a lot of what the Data Science team is doing is on the forefront of productionalized ML, much of my work doesn’t have a tried-and-true way or easy StackOverflow answers. I’m constantly researching and experimenting. Recently, I’ve also begun checking out the machine learning subreddit and googling other general topics to see if any new developments have occurred in the field that I might find useful. When I really get stuck on a problem, I prefer to step away from my computer and get some fresh air. 

 
 

Walking meetings are a common occurrence on the Data Science team. I return from the walk feeling refreshed and often have a new idea of how to attack the problem. After a busy morning of problem solving, I can finally eat the lunch I have been hungrily thinking about!

My afternoons consist of continuing to make progress on my research or having meetings with other MLEs. They might introduce me to a new part of our tech stack or I’ll update them with my current progress on the project. If I don’t have a scheduled meeting and hit a snag I can’t seem to figure out, I’ll usually message one of the engineers on Slack for help. They are quick to collaborate and offer a possible solution or point me in the right direction.

At the end of the day, I wrap up by updating my Trello with the day’s progress and add any new tasks to the board that might have arisen. I check my inbox for remaining emails I missed, push and back up any changes to files, tidy my desk, and clock out. My drive home is a little longer than the morning commute, but still not bad in comparison to the traffic I see when I drive over the highway.

I like to unwind after a long day by hanging with my roommates, playing Rocket League, or going to visit my girlfriend. While I don’t have the most successful track record, I also try to work out a couple times per week. I wish I would spend more time playing guitar or producing music — after all, my roommates did build that recording studio in the basement and it would be silly of me not to use it! I finish the day, finally, and am in bed around 10:30p to get adequate sleep and be ready to dig in to the next day.

I’ve now completed the first month of my internship and so far I’ve been very happy with my experience as an MLE Intern at The General. The Data Science team is smaller, affording me the capacity to work on new and cutting-edge problems that will have a positive impact on the business and make our customers’ lives easier. I love walking in each day knowing I’ll be working with passionate and motivated colleagues, have the opportunity to work on what’s never been done before, and discover exciting solutions. I’m not just creating ripples in an existing repository; I’m making waves in the data lake.

 
James Bertrand