Wednesday, August 26, 2015

Astronomer to Data Scientist, Three Years Later

In August 2012, I began my first job as a data scientist. I handed my completed dissertation to my committee on a Friday and the following Monday I started work. Leaving academia (and astronomy) was not an easy decision for me. I remember starting that first day thinking: "Well... if this doesn't work out, I guess I'll reapply for post-docs again next year." It ended up working out better than I could have imagined. I wrote several posts about this transition during my first year working as a data scientist, but I thought I'd reflect and talk about what it has been like working as a data scientist, now that I am further along in my career.

Much of what I said in my post Astronomy vs Data Science continues to hold true today, and so below I will simply add some new perspectives now that I have a bit more experience in the field. 

The Pace
My job continues to be very interesting and challenging.  I am always learning new things and working on new problems. The pace of projects is much faster than research (days/weeks instead of months/years).  This means that I can't dive as deeply into problems as I could in academia (con), but that I rarely get sick of projects or feel like they are dragging on too long (pro).  The work/analysis I do is (in general) not as mathematically complex as that which I did for my dissertation research, however I have learned a wider rage of technical skills (software engineering, visualizations, modeling, natural language processing, machine learning, time-series analysis, financial analysis) that I doubt I would have learned if I had continued in astronomy. I am also learning other types of skills like project management, team management, team building, communication, and business intelligence.

"In physics, I did one really interesting analysis every year. In data science, I do one kinda interesting analysis every day. Though the individual problems aren't as deep, my overall intellectual engagement is higher." -- Andre Bach, Data Scientist, Uber

"I like the fast pace. I get to create new models roughly once a quarter, and my colleagues typically adopt the resulting predictions immediately. I feel like I'm making new, tangible contributions all the time. In academia I felt like my preferred pace was many times faster than the typical paper or grant cycle, so I definitely prefer it this way." -- Will High, Senior Data Scientist, Netflix

My Day to Day
As I advance to more senior roles, I am surprised at how much opportunity there is for me to continue to perform analysis and do technical work. I would say that 60-70% of my days are spent working on data problems. The rest of my time I am in meetings, helping teammates, writing reports/talks, or working on recruiting/interviewing new teammates. I really like the fact that tech companies typically have an individual contributor career track where you can continue to move up within the company without needing to be a manager.

The Focus
In industry everyone is focused on improving the customer experience, and ultimately the bottom line.  This means that I only work on projects that are aligned with this goal. The downside is that I can't explore anything and everything I find interesting. The upside is that I am rarely in a situation where I am stuck spinning my wheels (an experience I had frequently in academia). Overall, I find it very rewarding to have this focus; To know that almost everything I work on will be immediately applied and implemented; To see my work making real changes to the business and improving the experience of our customers on very fast timelines. It's incredibly motivating.

The People
Some of my closest friends are people I met during graduate school. I was concerned about leaving academia and leaving all the awesome people. However, I have found that my coworkers are just as fun, interesting, and smart in industry as they were in academia. In addition, I have found industry to be a more collaborative and supporitive environment than academia, however I recognize that this probably has more to do with my particularly companies/institutions than industry vs. academia. I also am pleasantly surprised at how I have continued to be involved in the astronomy community even though I am no longer an academic. I was very worried that people within astronomy would view me negatively because I went to industry, however I have not found this to be the case.  In fact I still get invited to give talks (about data/career stuff), and I have attended astronomy conferences.  

"I get to work directly with extremely bright professionals who specialize in completely different disciplines. My colleagues are awesome." -- Will High, Senior Data Scientist, Netflix

The Pay
In San Francisco entry-level data scientists (coming from masters or PhD programs) make between $75k and $125k base salary. More established companies (larger/publicly traded) tend to pay more, have more benefits and perks, but offer less equity (stock in the company).  However this stock (if the company is post-IPO) can be sold and become real money. Less established companies (pre-IPO/start-ups) tend to pay less and have less perks, but give you more equity. However that equity might never end up being worth anything if the company fails.

I recently went through a job search and my base salary offers were between $130k and $170k. Assuming that the average entry-level data science job offer is ~$100k, and the average senior-level data science job offer is ~$150k, this is an increase in salary of 50% in three years.

On top of base salary most tech companies will issue stock grants. The value of these grants vary dramatically but my data science friends have gotten initial stock grants valued as low as $60k and as high as $700k (with a 4 year vesting schedule). Some companies will give employees annual grants/bonuses on top of the initial stock grant.

Overall I've been very impressed by the compensation in the tech industry. Especially considering I continue to have a great work-life balance (my typical hours are 9-6 and I rarely work weekends).

Turnover
The typical tech worker changes jobs every two years. This was scary for me coming from academia where the goal is to be a tenure-track professor and have the same job for most of my life. However now that I am in the industry I realize that people are moving around a lot because they want to keep learning new skills and because the job market is such that you are constantly being approached by new companies offering you exciting new job opportunities (I literally get 2-5 people a week reaching out to me on LinkedIn asking me if I am interested in a new job). However, there are plenty of people who stay in the same tech job for much longer than two years. Personally, my motivation for changing jobs was not because I was unhappy at my current role but more because I've been excited by a new opportunity.

This transition is now a well-trodden path. There are a lot of us ex-astro/physics people who have transitioned into data science. We have a Facebook Group and we also regularly meet-up for happy hours in San Francisco and San Jose.  If you have more questions about transitioning from academia to data science feel free to contact me or ask questions in the comments section below!  We can help connect you with other people who have made this transition or help you break into the industry.