Metis bootcamp graduate Andre Gatorano is now a lead Data Scientist at Blitsy, where he works on a number of projects aimed at helping both the company and customer maximize their digital experiences on the site. I recently caught up with Andre to talk about his multi-purpose role, to discuss what tools he's using on the job, and much more.
Tell me about your background. How did you become interested in data science?
I had experienced an enormous amount of change in my interests and career pursuits before reaching data science. Upon reflection, the only thing holding them all together was my interest in the problems that data science gives you the tools to solve. I started as a bioengineer who ended up working in a genomic cancer research lab. R was beautiful and enticing. The data felt powerful. I pursued bioinformatics full time. I then was pulled into working for the San Diego Supercomputer Center. There I analyzed and visualized huge datasets involving international torrent use (cool, right!?). All of this happened before I become intrigued by public health. My senior thesis used a logistic regression and neural network to estimate the severity of the Yellow Fever epidemic. At this time, I still did not know what data science was, but even so, I guess you could say I was always interested in data science.
Describe your current role. What do you like about it? What are some challenges?
I am currently a lead Data Scientist at Blitsy, Inc. I have no specific goal other than to help make the entire site is responsive to the user and to help the company better understand its customer. I am building out the recommendation system of the blitsy.com platform, which means making sure everything presented to the user comes from logic and is better suited to the specific user. I am also running classifiers to predict if someone will drop a cart, and am identifying individuals to market specific items to. A lot of hats. I really enjoy how much breathing room I am given to explore powerful tools. Even better is how my work will be a major piece of the website's behavior and user experience. Eventually, everything you see on the site will be generated from my work! I have help from people who understand the industry very deeply, and I am learning how nothing is as cut-and-dry in the real world of data science - especially for a startup. However, good data can be hard to come by. I am having to work hard to find good data to work with, and am transforming what I do have into something more valuable. But this is an exciting opportunity in developing a strong data infrastructure in the company.
In your current role, what aspects of data science are you using regularly?
I am using PySpark often to run matrix factorization algorithms, along with column cosine similarities. I am using classifiers to a lesser extent when the data is available and the motivation is clear. In the beginning, much of my time was spent running through a lot of statistics about customer behavior and using clustering algorithms on users. I have also used Gensim and word processing quite a bit. Statistics and problem solving are my best friends. I am fortunate to be using data science and big data tools regularly.
Do you think the projects you did at Metis had a direct impact on your finding a job after graduation?
Absolutely. I was amazed at how deeply knowledgeable I was in the interviewing process. At times it was clear I knew much more than I was anticipated to know and my breadth of knowledge was greatly appreciated. For most data scientists who are through a Ph.D. or are self-taught, they may not know how to do much natural language processing. One may know how to create powerful classifiers, but not how to scrape websites. I had such well-rounded experiences that employers were often curious as to how I could know so much. Make sure you blog. Blog, even just for yourself. Give yourself the chance to process what you are learning and identify how to effectively convey your knowledge. Also, it gets you jobs pretty nicely. ;)
What would you say to a current Metis applicant? What should they be prepared for? What can they expect from the bootcamp and the overall experience?
Everyone is different. Don’t worry about your background. Even the people with prior experience have huge holes and blind spots. One area you may feel behind, the next everyone is on equal footing. And in another you will feel like you are soaking things up better than you thought possible. I recommend spending time in the areas you don't feel comfortable in. Once you are working, you don't want to feel intimidated by an area of data science. You can always learn more about something once you are on the job. The real value is the confidence gained by spending time to get over the initial fear. No task or algorithm is that crazy. It may seem like a mountain at first. Once a little time has been spent, you will find how reasonable most things are to be effective in. Take a deep breath. Focus on understanding more than completion. Set a goal for yourself in a project. Say, "in this project, I am going to get good at natural language processing, and especially LDA." Learn about those things and craft a project around them. Too often people are consumed by how cool or grand a project can sound. In the end, the project is a launching point in learning. By setting reasonable and pointed goals, you can make sure to have a sound understanding of most of the topics by the end. Metis will be there for you. This involves the lecturers, the staff, and your peers. You don't have to do things alone. Be steadfast, but not completely unyielding. Know your weaknesses and get help.