I’ve decided to move on from both Nielsen and data science and start a new role in engineering. This is a short post with some reflection.
My new role
I’m going to be working as a software / machine learning engineer at Tempus. Tempus works in “precision medicine,” which I understand to be individualized healthcare that relies on ML on DNA, RNA, and protein sequences. My team, Tempus Algos, ships these ML algorithms to doctors.
They shipped their first product last year, and since then they’ve launched a few more. The goal is to launch many more algorithms over the next few years. This made it feel like an exciting time to join this team specifically.
Reading and writing less
This is why I’ve been less active writing blog posts this year. Job searching drained my energy. I had little motivation to read anything after prepping for interviews and constantly thinking about what the future holds.
Given this, I’m quite grateful for my reading club with my friend Judah, which forced me to read a paper a week. This is a minimum that I’m satisfied with, even though last year I was reading more.
And I miss reading, to be sure. I went to a coffee shop patio a couple days ago to read a paper for the first time, and it was refreshing; it reminded me of the pre-COVID before times, when I would sit in a coffee shop for an entire afternoon to do nothing but read.
My hope is that I’m able to ramp back up again. I’ve read a few papers from FAccT and have my eyes on another couple that seem interesting. I’m looking forward to reading a lot from CHI, which is happening in early May.
Leaving Nielsen
There was no single event that made me decide to leave Nielsen. My shields have been down since six months in (February 2019—an eternity ago!), when I was placed on a terrible project to translate SAS scripts into Python. I was able to work on something else after a couple of months, thanks to help from some mentors. But since then, I’ve one foot out the door and was casually interviewing every couple of months.
Why now, though?
Still, nothing specific. I liked my manager. I liked my team. I’ve met and worked with lots of great people. I was working on some of the most interesting work at the company. But I still felt like I was missing something, which turned out to be a combination of impact and mentorship
Mentorship: I felt like I was missing working with talented engineers who could teach me better practices. I knew that this was out there—I got it as my internship at Qualtrics—and I’d been leaning on new-grad-level engineering skills for too long. Talented engineers exist at Nielsen, but data science being so big, I never got to work with them.
Impact: in my work on a research team, I felt like I never was able to see the impact of what I did. Two of the models I built were shipped—but I use the passive voice because a different team shipped them. I was only lightly involved in the process, and I have no insight into how this affected our clients or our business.
Relatedly, I realized that there were so many things out of my awareness, leave alone control, that decided whether my projects would matter or not. I appreciate that this is the reality of a lot of research work, but it’s not what I felt I needed early in my career.
Finally, the logistics (and luck) also played an undeniable part: the right opportunity came up. I got more recruiter messages in January than in most of 2020, thanks to companies increasing their hiring. The further along I got in the interview process, the clearer it was that it was time to move on.
Why I’m stepping away from data science
I’ve gotten this question from friends, coworkers, and family, but I don’t personally think this distinction meaningful. This role is still a part of the overall data science lifecycle, just downstream of the actual model development (which, to be honest, was only a small part of what I did as an actual data scientist). The title “software engineer” matters less to me.
I believe that:
- right now, it’s important for me to focus on my engineering skills
- it’s particularly important to get experience shipping models and driving business impact
- I will learn a lot working at a small company regardless of the title
- I can still return to a data science role in the future
So I’m not beat up on “leaving” data science. I don’t think that this door is closed. Working as an MLE will certainly make me more well-rounded, and likely more marketable in the future.
Closing thoughts
Above all, I’m excited.
I spent a lot of energy on the job search. I dedicated lots of energy to deciding if particular positions were right for me, preparing for interviews, and deciding on a next step. I’m better off as a result, but I’m very glad that it’s over.
Right now, I’m on vacation for another week to recharge. I’m working on completing a living dex in Pokemon HOME, reading more, and going for daily walks. I know that starting a new job will be filled with new challenges, but I’m looking forward to the opportunity.