The Five Stages of Interview Grief
Unicorns aren't real. Unicorns aren't real.
It’s been over four months since I left my job to take a break, and with a fresh perspective and optimistic outlook, I’ve started to turn the crank on interviewing again. After casting a relatively wide net for prospective employers, I’ve had a lot of productive conversations and feel mostly in the right frame of mind to be successful at the end of it all.
But even after approaching the process with a confident attitude, there’s nothing like interviewing for a new job to make you second-guess your abilities.
True unicorns are rare, and the rest of us mere mortals experience interviews that are designed to test the magnitude of our deficiencies as much as our skill sets. My career path has been a little different than most other data leaders at my level, because technically I have never been a Data Scientist (individual contributor). Gasp! It’s true, almost all of my hands-on technical knowledge (mostly batch processing with C++ and python) and applied statistics programming (RooStats/TMVA) was learned when I worked as a practicing research scientist in academia. I’ve been in management for over 7 years since then, and if you sat me down to do EDA with Spark or even a complicated SQL query on the fly, I’d be able to muddle through it, but at a fairly disappointing clip.
It may not come as a surprise that I don’t believe there is such a thing as a “standard career path” for data practitioners. In fact, I think more often than not, the experience requirements1 employers rely on in order to select candidates for a particular role is, in reality, an amalgamation of the best traits of everyone who has been in that role, and is actually just an industry-wide representation of the perfect employee. As hiring managers, we tend to be fairly risk-averse, mainly because we usually have only one shot to get the right person into a role. The stakes are high, and so unicorn-hunting becomes the default.
In any case, on this side of the interview table, I have to put up with it and play the game. Back to the books, I’m running through exercises to refresh my memory of certain topics and terminology. I am fortunate to have the time to study, which many do not, and I know this puts me at an unfair advantage. But even with all the time in the world, I intentionally chose the management path over the IC path, a choice that has meant there are certain experiences I just won’t have as long as I keep going on this road. Certain experiences that, unfortunately, prospective employers have been quick to point out as missing from my resume.
I have to let it go, but not without first going through a few stages of (interview) grief:
Denial: Practically perfect in every way.
“What are you talking about, of course I could do what anyone on my team does.”
Anger: Time to get extremely defensive.
“You don’t think I’m technical enough?? I’ll show you technical! I was working with petabyte-scale data when you were still in high school! I’ll…I’ll…get my old research code and stick it on github for all to see!” [dusts off 12-year-old laptop].2
Bargaining: Trying to get a callback by fitting a square peg into a round hole
“Wait wait, don’t dismiss me, I’m capable learning anything, I could totally do whatever you have in your job description even though it’s not really something that interests me.”
Depression: The scotch and sweatpants stage
Damn, hello imposter syndrome.
Acceptance: Oh yeah, I forgot. I’m awesome.
For me, acceptance came when I reminded myself that every time I have felt like my job really should have been two distinct roles, I’ve also felt like I did each of them half-assed. It’s reinforced my already strong belief in the concept of a “technical lead track” for ICs as a different yet equally fruitful career path than the management track.
I was given some great advice once, that I should always take the time to write the things that I’m proud of accomplishing into a living “hype doc.” It’s important to remind ourselves what makes us great, and there’s no shame in that. It’s especially true when interviewing. I’m very confident in my ability to build great teams who build great things, and my future employer will value that in me.
Now if you will excuse me, I’m going to dive back in to Mode’s tutorial so that I can refresh my query skills and remember how much I dislike SQL syntax.
Epilogue: A relevant anecdote
After my first year of graduate school, we were required to take a two-day, comprehensive qualifying exam. No notes or books were allowed, only a pencil, blank sheets of paper, and our brains. The scope of the test covered anything we had ever learned about physics in undergrad or graduate school up until that point. If we didn’t pass, we were kicked out of the program, regardless of whether or not we were already doing research. It was a make-or-break rite of passage that many schools have since done away with due to its abject cruelty. Senior graduate students, postdocs and long-time professors would say that in those two days, you would know more textbook physics than you ever did before or ever would again in your career.
On one particular study day a few weeks before the exam, some fellow students and I gathered in a tenured professor’s office to gain a better understanding of a subject that he literally wrote the book on. But when we pressed him for details on how to solve one problem in particular, he said that we were on our own to figure it out because he couldn’t remember exactly how it worked.
“Well,” he said, relaxing back in his chair, “I’m not the one taking the test.”
Who hasn’t seen that job description: “Entry-Level Data Scientist: We want to hear from YOU if you have a PhD in a quantitative field and 3-5 yrs professional work experience in a data science/ML role, are an expert in statistical and experimental methods, has deployed predictive models and machine learning algorithms to production and is well-versed in Clustering, Regression, NLP methods etc, has demonstrable experience developing company-wide metrics with extreme data visualization skills and has produced actionable insights with exec-level visibility, fluent in Python/Java/Go, can do SQL in your sleep and has hands-on experience with Scala, Hadoop, Hive, Spark, Kafka, Kubernetes, Airflow, TensorFlow, D3 and Django. Nice to have: Domain experience in our niche market.”
Well, I did not put my code on github, because it’s old and gross and anyway this is why I’m a manager now, but yes I did find the power cable for my 12-year-old Dell Lenovo, and yes I did manage to boot it up (aww Ubuntu 10.04) and yes, it did feel like digging up a sunken treasure full of gems, for exampe:
> less ChangeLog
2012-05-30 <ethompso>
* getting boostedgoos back into SVNINST after svn disaster of 2012
* Tagging version BoostedGOOS-00-00-05I cannot for the life of me remember what the “svn disaster of 2012” was but it must have been awful.
Ah, the tenured professor who has forgotten the details, great image. The industry equivalent is when the senior data lead -- that grilled you in the interview -- on your first day says, "Yeah, so this is terrible practice, but here's what you have to do to get going..."
Emily, I'm really impressed at your ability write both clearly and with tons of entertainment value! I find myself nodding along every time I read one of your posts.
One piece of advice I want to give you as you look for your next great job, if you don't mind, is this: don't worry about the things you can't do. Focus on the things that you can. How is it you deliver value to orgs? What difference do you make in the careers of your people, the missions of your stakeholders and the outcomes of the company? That's what should be coming across in interviews and if that is what the company values...there's no fit.