This post was written in partnership with SwitchUp. You can find our recent guest post about students and alumni using data for social good on SwitchUp's site here.
Marcus Carney’s career has spanned many different roles. After serving as an Airborne Infantryman in the United States Army, he started school at Ohio University to earn a Bachelor’s of Business Administration in Finance. He then kicked off a career in the financial sector, as an Analyst for companies like JPMorgan Chase and Credit Suisse.
Working as an analyst gave Marcus a first-hand look at how big data and machine learning were starting to change the world of finance. In an effort to get ahead of the curve, he started to teach himself Python and Machine Learning with the help of online courses.
Once he had learned the basics, he decided that a bootcamp would offer him the best balance between cost, flexibility, and content so that he could transition into a full-time data science role. When it came time to compare programs, Metis stood out among the rest.
“Once I decided to go down (the bootcamp) route, it was easy to settle on Metis. It was accredited, was focused solely on data science, and had a robust alumni network and career support. I looked into several other options online, but Metis stood out above all,” he explains.
In an interview with SwitchUp, Marcus talks more about his path to a data science career, his experience at Metis, and his new role as a Data Scientist at CKM Advisors.
SwitchUp: You started your career in the United States Army, and later worked as a Financial Analyst. What made you decide to learn Data Science?
The growing realization of how quickly the world is changing. In my role as a financial analyst, I would consistently hear other analysts and company executives talk about the opportunity and potential of machine learning, big data, etc. for finance specifically and the business environment overall. The post-financial crisis focus on regulation expense management has led most large financial institutions to aggressively reduce headcount and automate functions, and the scale of data available to analysts is far outpacing the capabilities of more commonly-used business intelligence applications. So when I looked at the trajectory of finance jobs vs. something like data science, I thought it best to get ahead of the curve.
How did you choose to attend Metis? What was your process to research bootcamps?
Once I decided to pursue data science, I looked at several options – masters degree, bootcamp, self-learning, etc. I taught myself Python and took the Coursera machine learning course, but was also working full-time so couldn't devote as much time as I would have liked to learning more. Going to a bootcamp seemed to be the best balance between cost, flexibility, and learning content. Once I decided to go down that route, it was easy to settle on Metis; it was accredited, was focused solely on data science, and had a robust alumni network and career support. I looked into several other options online, but Metis stood out above all.
What skills were you hoping to build at a data science bootcamp?
When I started, I wanted more exposure to machine learning and neural networks – the cool stuff. Though I still am very much interested in these topics, through the bootcamp I was introduced to more nuanced topics of interest, like writing efficient code, memory optimization, effective visualizations, and a bunch of vital but less-publicized aspects of a data scientist's toolkit.
Tell us about the learning environment at Metis. What was the curriculum and classroom instruction like?
The curriculum was wide-ranging and touched on many of the ideas, frameworks, and tools data scientists use – data acquisition and cleaning, Pandas, supervised and unsupervised machine learning, structured and unstructured databases, etc. Classroom instruction was lecture-based in the mornings, generally, and in the afternoons you would work on your projects and the instructors would be available for questions and more personalized help. It was nice to have the balance of both guided learning and self-instruction, especially to be able to focus on topics of interest to you individually.
As someone who already had some experience with Data Analysis, what was your biggest challenge in the program?
Figuring out how to make Python do what I wanted it to! After over 3 years of using SQL and Excel, I was fairly adept at running relatively complex analysis with those tools, and though I knew what it was I had to do (aggregations, regressions, etc.), at times it was frustrating to re-learn how to do it all in Python. It was absolutely worth the frustration, though, as I can run analysis orders of magnitude more complex in size and scale than previously.
You worked on several interesting projects while at Metis. Could you tell me about one of them?
My favorite project involved natural language analysis on emoji (see project slides here and accompanying blog post here). The week before our project started, I was watching an episode of South Park in which an internet troll was harassing students on an online message board. One of the girls in class was analyzing emoji usage to try to determine who the troll was, and I thought, huh... I wonder if you can actually do that? I pulled over 1 million tweets containing emoji from Twitter and ran both text and emoji through a word vector model, which can measure semantic word usage. I developed a simple model of emoji "sophistication" using the word vectors to measure the similarity between text and emoji and variety of emoji used. It worked well, and I was able to classify tweets in which people used emoji to tell stories or embellish their tweets with remarkable accuracy.
What was your job search like? How did find your current role?
Job searches are never fun, but it wasn't anywhere near as onerous as the searches of some of my non-data-science friends. Thankfully, the skills I learned at Metis are in high demand, and with both my previous experience as a financial analyst and the fact I am a veteran, it was a bit easier to get interviews. They told us the whole process would take 2-3 months, from initial interviews to offers, and that was accurate. It also helped to start the job search in January, when most people are coming back from the holidays and companies are preparing for the new year.
I actually found my current role through Metis; at the end of the bootcamp, they have a "Career Day," in which you present your final project to a room full of alumni and recruiters and have the opportunity to network afterward. I spoke with the representatives from CKM Advisors there, and the rest is history.
Now that you work as a Data Scientist at CKM advisors, what is your day-to-day role like?
It varies; I've been lucky to have worked on some interesting client projects, so on any given day I might be attending meetings, running analysis, acquiring data, or any combination thereof. When there's downtime, I'll take online courses or read up on new tools and techniques – there's no shortage of free resources out there for the data science community to learn and new things are constantly being developed.
How do you use the skills learned at Metis in your new role?
The project-based nature of the curriculum gives you great practice in framing problems and developing strategies to attack them, which I use almost every day due to the nature of our work conducting analysis and developing data-driven applications for our clients. I've applied natural language processing and machine learning to identify patterns and trends not evident in simple aggregations or number-crunching. As we start to productionize code, the fundamentals of computer science and code efficiency are starting to come in handy. And of course one of the most valuable skills I learned at Metis was data acquisition and cleaning – most data scientists I know spend a bulk of their time cleaning and sense-checking data, and in making you acquire and clean your own data, Metis implicitly teaches those skills in addition to the more overt curriculum items.
What are your career goals going forward?
Generally, to gain exposure to different projects and tools to further deepen my data science and project experience. The overall field of data science is still relatively new (and changing rapidly), and CKM itself is growing very quickly as a company, so I see plenty of opportunity for this.
What advice do you have for people who are interested in attending a data science bootcamp?
First, I would advise doing research and speaking with bootcamp alumni to find the best place for you. There has been an explosion of tech bootcamps in recent years, and the quality varies, so make sure you're going to spend your time and money on something worthwhile that will take you where you want to go. A great way to do this is to reach out to alumni on social media (like LinkedIn), and ask them about their experiences, and attend events if the bootcamp hosts any to get a feel for the place. Another specific thing to look at is career support – one of the things that drew me to Metis over other bootcamps is the very strong career support network they offer.
Otherwise, one of the best pieces of advice I received was from an instructor early on in the course. What encompasses "data science" is so varied and nebulous that many bootcamps touch on several different topics quickly, and may not go into the depth some people want or expect. This instructor told us to think of the bootcamp not like a class (where you have to learn everything at great depth), but more like a music festival. You wouldn't try to see every band at a music festival – you'd pick the times and stages of the bands you want to see, maybe go check out some new ones in downtime, etc. Much like that, since it's not possible to learn everything there is to know about data science from a bootcamp, don't try – pick a few topics you're really interested in and spend a disproportionate time focusing on those rather than trying to wear yourself out learning EVERYTHING.
Check out what other students and alumni have to say about Metis on SwitchUp’s Metis reviews page.