I recently read an article on the state of data science and advice on how to enter and succeed in the field which really hit home. A large chunk of the article, Data science is different now by Vicki Boykis, is devoted to the oversaturation of the market for data scientists (especially when that is the specific job title), misconceptions about the daily duties of the job, and how those things relate to her path. I can’t yet speak to any misconceptions about daily duties as a data scientist, in part to not having that title and in part to having just started my first job in the tech industry. However, I can relay my experiences with the oversaturation of the market and how I see my own way forward.
In late 2016, I made the decision to apply to graduate school. At the time I was working for the NBA and happy with my job, but the work still had a seasonal aspect to it and the path for upward mobility was unclear. I had my undergraduate degree in math and an interest in sports statistics, so I looked at places that I could potentially work for the basketball team and learn more methods and skills that could help with sports analytics. Of course, like so many others, I saw the articles celebrating the rise of data science and while there were few programs with that distinct title, I applied to those that existed or sounded close enough. I ended up at the University of San Francisco in the Masters in Analytics program, which changed to the Masters in Data Science program halfway through my year there. Around March it came time to start applying for jobs, and once May hit, the job hunt was in full swing for basically all of us in the program - except the lucky ones who already had jobs. Searching for jobs in the NBA tends to be fairly tough and competitive, so I’d dealt with disappointment before and thought I was decently equipped for this round of hunting.
I was wrong.
Nearly seven months, over two hundred and twenty five job applications (though 38 were LinkedIn Easy Apply), numerous iterations of my resume, plentiful networking events and coffee chats and phone calls, five on-sites, and one prolonged visit home later, I had a job. I wasn’t the last person in my graduating class of around 75-80 to get a job, but I was one of them. It was an arduous process that I learned a lot from, including my weaknesses and general interview know-how, and by the end of it I was not even a “data scientist” in title.
The application process was excruciating at times. Cover letter after cover letter. Paralysis by analysis of how I should spend my time (more applications? personal projects? interview prep? Kaggle competitions?). I looked up every company I saw anywhere - on ads, on people’s backpacks or sweatshirts, their office signs - to see if they had openings. It became clear over time that my prior experience in basketball, and definitely not tech, was both a conversation starter and ender. I think one guy even confused basketball with mobile gaming and told me to look for jobs there. Plenty of email rejections and some over the phone. There were a few especially tough ones, such as twice getting beaten out for an offer by good friends.
Eventually, I broke through. But a job is a job, and I’m quite happy with it. I might not be a “data scientist” or a “machine learning engineer”, but I’m in a data job and it’s great. It’s a cool company and my coworkers have been great so far, especially the dogs people bring in. I think I’ve already started learning a couple of the lessons in Boykis’ article, namely SQL and writing efficient and clear code. I knew SQL a bit going into my program, learned more of it there, and generally thought I knew it. However, I had never seen a query over forty or fifty lines or really used the WITH statement before. I’ve now seen both of those, though I still haven’t done much in the way of database management.
Writing clear and efficient code is a huge thing, and considering I’m learning built-in and/or basic Python functionality every few days, it’s definitely still a work in progress. However, I’ve re-factored code from past data challenges to speed it up, learned how to (safely) automate processes, and picked up tips about general writing and testing. I think I’m doing a pretty decent job of leaving comments and documenting any code that isn’t purely scratch work. I am definitely interested in learning more about the “hardcore computer science” side of things and how that affects my own interests, projects, and work.
This leads into my final point, which relates to my first article about goals for 2019 and connects back to Boykis’ suggestions. There are a few things connecting a few of my goals and act as a bit of a foundation for them through this connectedness by providing a boost in efficiency and productivity. Learning these underlying skills and acknowledging them as part of the process will help me not only with these projects but with plenty of other things moving forward. They boil down to becoming adept with the command line and terminal in addition to becoming a power user of either Sublime Text or VS Code (I guess I need to resolve that first) to help streamline my workflow, at least when I’m not using Jupyter notebooks.
I believe focusing on these foundational skills and tools will help me achieve my listed goals, especially as I develop some more detailed plans around accomplishing them. Knowing how to operate and move around in both the terminal and a code editor will improve my work performance and contribute to my skills for a long time after any of these goals are met. Hopefully, that makes continuing my career a lot easier than starting it.
Comments