On Being a Data Science Intern

A snapshot of a summer at Knewton

First, a little background on me. At the start of my internship, I had just finished my second year as a PhD student in statistics at the University of Michigan. My research is split between modifying sequential decision making algorithms for uses in education and poking around in the nascent field of fairness in algorithms.

The Application Process

I can’t say if I over-prepared or prepared just enough, but I prepared a lot for the somewhat-surprisingly-long interview process that Knewton has for data science interns. There were three steps to my application: an initial phone screen, a data science project, and a final interview through video chat.

  • The initial phone screen was maybe an hour. We talked about my interests in Knewton, what I had already done, and why I was a good fit, along with a smattering of stats-related questions.
  • The project took me longer than I had imagined. This was mainly due to having to reacquaint myself with Python (since the data science team at Knewton uses Python) after programming predominantly in R for grad school. In all, working between my own classes and teaching, it took about a week. It is possible to spend a lot of time on the project, not get the internship, and be bitter about it. Instead, I approached it as an opportunity to improve my coding and work on an interesting problem that wasn’t so cut and dry as what I had encountered in school.
  • The final interview was three hours long. It covered statistics, programing, a general conversation about Knewton, my interests, and questions I had about the company/internship. I spent a lot of time preparing for the programming portion, and no time preparing for the stats portion. I recommend not doing that. Spend time on stats.

First Impressions

Knewton was flexible in my start date and allowed me to begin a month before the other six interns, to accommodate plans I had in August, which made it more like I was a new hire than an intern.

During on-boarding, I felt I was being fully integrated into the six-person data science team. There is a lot to learn to get up to speed, but fortunately everyone was very available to answer my perpetual questions… even if it was the same question more than once (pro-tip: write everything down). I was also given a mentor, who was a constant source of support for me.

Integrating into Knewton

Being so small, the data science team was vocal to me about valuing outside perspectives. I was immediately included in all team meetings, and encouraged to share my opinions. I wasn’t expecting to feel so involved in the current workings and future plans, but I was, and it has been amazing experience.

For example, in a meeting meant as a Q&A for another team at Knewton, I often interjected my own questions and thoughts because I was curious about how things worked and why certain decisions were made. Afterwards I felt as though I may have overstepped and taken up valuable time. To my surprise, my manager found me afterwards to tell me they appreciated my input and the questions I raised, and said I should to continue to contribute.


My mentor did a great job of choosing my first project as a way to acclimate me to Knewton’s data and code flow. It was much easier to get used to the lingo and understand the current algorithms for student proficiency and the recommender model once I had a purpose.

The first project was to identify and write metrics that could be used in a weekly report about student experience in Knewton’s alta product. This was as much a lesson in planning as it was in coding. If you’re coming from academia (as I am) where you are in a single person lab, and collaboration means giving your advisor a quick update every week or so, then it is quite another thing to fully document what you are doing and why, so that someone with no knowledge can understand the work you leave.

Independent problem solving, collaborative critique

The main project assigned to me once I was up to speed was incredibly interesting and also incredibly open-ended. Which of course is the the best kind of project.

Knewton’s alta is a mastery-based learning platform. It is important to quantitatively define mastery, proficiency, and completion, such that students are getting the most from their adaptive experience while also having a positive one. With data collected from previous semesters, I was tasked with testing whether the current definition for completion of an assignment, in terms of proficiency and future performance, could be improved upon.

Currently, literature on mastery learning has no agreed upon mathematical definition of ‘mastery’ or ‘proficiency’, so I had a lot of freedom to play with. I was able to sequester myself to read papers, dive into the data, and come up with several ideas. Once surfacing, the team would meet and I would refine my ideas based on their feedback. After a couple iterations of this, I did what my mentor had considered potentially impossible, and produced some reasonable conclusions. The word ‘proud’ was mentioned. It was a good moment for me.

Extra Curriculars

One of my favorite things about working at Knewton is how everyone is truly dedicated to helping students learn. To that end, there are several groups one can join within the company that meet and talk about current research with an aim to improve all aspects of Knewton.

I enjoyed the pedagogy working group the most. During my time at Knewton we read papers on things such as collaborative learning, gaining lasting learning through revision, and soliciting student explanations to deepen understanding. This was also an opportunity for me to lead a discussion and share research I had learned at Michigan.


The culture at Knewton is truly wonderful. In my opinion it is the perfect size company for an intern. Knewton is large enough and established enough to teach you good practices and give you great mentors in several different areas, but small enough that you can still meet everyone by eating lunch in the kitchen and can be involved with the whole company if you so desire.

Not to mention the Knewton anniversary party, the frequent afterwork “tea times” (these are just more parties), free lunch and perpetual snacks, and all other amenities you would expect at a tech startup.

Why should you want to intern at Knewton?

I don’t think I would have been given such an interesting and important project at a larger company. Being trusted, given independence, having my opinions heard and acted upon, and being provided constructive feedback throughout the process: that is what made this summer so meaningful. Plus I’m much more adept at Pandas now.

For me the biggest benefit of interning at Knewton is that what I did matters. I wasn’t improving ad click rates or pushing money around in fancy ways. I was in a field I am passionate about working to improve student learning and experience.

Like what you read? Give Laura Niss a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.