How my Master’s got me to the forefront of Artificial Intelligence in 18 months
It’s hard to explain how I feel about Northwestern’s Masters in Predictive Analytics. When I graduated in June 2015, I was relieved I finished the course, but when I went looking for jobs, I didn’t know how to get started. Those first few months in the job market taught me a lot about analytics job market. It is hot — but only if you know what you want.
For simplicity, here are the 3 phases of my career over the last 18 months:
- Networked my way into a Fortune 50 Cable company business intelligence group.
- Standard application and interview with a Fortune 500 Media company’s brand new data science team.
- Jumped to a startup focused on Natural Language Processing and Artificial Intelligence.
When I started MSPA, I was part of the business analyst program at a Fortune 50 Cable company. As a business associate, I learned a lot about budgeting and planning, but we were always making assumptions about the business. The closest we got modeling the behavior of our business was a linear regression on less than 5 variables. It was a $10 Billion dollar group — I knew there was a better way to get answers to our question. I found the world of predictive analytics based off a google search and knew that is what I wanted to do.
Back to my story, June 2015…
I had just finished my thesis in about one semester (highly recommend wrapping it up very quickly). I was working at one of the three largest cable companies in America working primarily out of Philadelphia. The job was a lucky break while I was in MSPA because I got to meet “real” data scientists. At that time, I thought about data science as a technical expertise as oppose to a business-first mindset focused on understanding the broader picture of how data science could solve problems. Call it misguided, but that focus on technical skills helped me narrow down the job search quickly to technical roles in data science. I also needed to move to New York for a relationship (who hasn’t?), so I started looking for technical data science jobs in new york. I submitted to a bunch of financial companies, a few media companies and startups. I got called back by a Fortune 500 media company and a startup. Interestingly, my financial background did not help me out at all. What might have helped was that I was strict with myself: 1) I did not look for data analyst positions because they did not seem technical enough, 2) I studied my course work easily 3 times before I interviewed.
You may be wondering why I didn’t like data analyst positions at the time. I have no problem with data analysts, in terms of impact, they have an immediate and large impact on the business because they use data to solve direct business problems. Data scientists, in my opinion, build complex models to solve questions we did not even know we had and have a higher risk of their work not affecting the business because it’s too exotic.
I also studied like a fiend for my interviews. I kept seeing roles that need Ph. D’s so I was worried I wouldn’t get another shot at my interviews. You really have to know the basics to get through the technical part of the interviews, that part is unavoidable.
I remember my interview — I was nervous because my future boss seemed very serious (little did I know he is never serious). He had a sophisticated Italian accent, that I spent the rest of the day imitating because I liked it so much. I talked about the MSPA program as heavy on code so it didn’t need in-person work. I also emphasized my experience working on remote study groups with people in multiple countries (true story). I also emphasized a bunch of repos I had on github. This helped a ton because we were able to pull up my work and talk through how I architected my code.
Shortly after, I started with the team. On my first day, my boss asked me to build a website. But I’m a data scientist. But we need a website. I built the website.
I had no idea where to start, but the MSPA program taught me how to be resilient from all the times I was stuck the night before on a homework and needed to use stackoverflow to solve a problem. I went to Stackoverflow and found a helpful post on how to get started in web development/cloud architecture. I basically followed it to the letter and about a month later, I had a functioning website and we were in business. Later on, we hired a web development director and I learned a lot more. Happy to report that while I did some things wrong, I also did a lot right and she was incredibly helpful in helping me grow. I also learned a lot about data visualization which was helpful to talk about in interviews.
In June 2016, I felt like I had grown a lot in the role, but I hadn’t built the really complex ensemble models I had seen when I got started in data science. I couldn’t force my business partners to accept a very complex solution when the gains were minor or non-existent. Hence, I started to look for more complex questions that needed complex answers. I also liked the engineering mentality of problem solving when I built the website. I decided to look for engineering/data science roles, where you could prototype and write production level code. That startup is Asapp, the folks there decided to give me a chance, and I’ll say I will always be thankful for the opportunity no matter which way things go.
At Asapp, I’m working on building complex machine learning models. Very complex. Complex to the point I still don’t always get parts of them, but I want to keep learning about them. I’m also writing a lot of server based code, and learning the basics of testing and scaling. A lot of our AI research is focused on neural networks in the language space, and while I can’t share what we’re doing now, I can say I love working here. Everyday, we’re talking about neural networks and cutting edge research in the field. Stuff that is maybe hours to days old from the best of the best in the industry. All awesome skills that give me the experience of being in a first class natural language processing team. When I originally asked what made me stand out, they said, “you really wanted to be an engineer. You really wanted it.”
See you in 6 months.