Why a Sommelier Chose Data Science
In my previous life, I was a sommelier. I went into the restaurant industry while supporting myself through my Bachelor’s degree and never found a way to leave, due in part to loyalty to my work but mainly to a crippling case of impostor syndrome creeping in any time I imagined myself working in a field like tech, engineering, or science. My degree was in math but I never felt like the type who belonged in academia; I constantly felt intimidated and like I was scrambling to be just good enough. Asking for help felt impossible, like admitting a weakness that someone could exploit.
Working in restaurants, however, developed my social skills, confidence, and adaptability in a way that college couldn’t. I worked from the ground up at the restaurant that became my home for five years, beginning as a server assistant and ending as a manager and wine professional. It was an incredibly challenging and transformative journey — I learned how to manage business finances but also to put out (actual) dumpster fires. I learned to write training proposals but also to communicate ideas and to stand up for myself. Every skill I absorbed from this journey has made me feel finally ready and competent enough to combine my analytical and communication abilities into pursuing the field of data science.
It seems like people are oftentimes categorized into the binary of being “feelings/perception” people or “logic/reason” people. Scientists, engineers, and analysts are the people who we call the latter. Many times, sommeliers and other beverage industry professionals are categorized as purely the former — they seem like magicians who pull arbitrary terms out of a hat to mystify the consumer. They curate experiences and elicit feelings by peppering in flowery language like “silky mouthfeel” or “structured tannins.” They give their guests the emotional satisfaction of being paid attention to (or the embarrassment of not knowing the difference between a cab and a pinot in front of their father-in-law). What does “rubber hose,” “pot-pourri,” or “crunchy texture” have to do with wine anyways and why does it all look like several raccoons in a trench coat walking into a country club? It feels like a theatrical performance, and for some, the illusion is so uncomfortable that it’s easier to just call them bad magicians. And just like with any industry, there are some genuinely bad magicians.
It’s easy to be a bad magician — all you have to do is build a vocabulary of buzzwords. But to be honest and genuine in this work is to find a way to marry the presentation and the logic into an experience that people can understand and feel comfortable around, an experience that is grounded in reason and showcases the analytical work that we put in every day to make it happen.
Wine doesn’t get the categorical and deductive credit that it deserves. After all, being a sommelier isn’t about throwing around olfactory adjectives and hoping something sticks. On the ground floor it is inventory management, profit-loss reporting, trend prediction, and consumer analysis. It is communication with distributors to curate a menu, with higher-ups to keep them abreast of financial details, with other staff to ensure that they have the necessary product and inventory knowledge to perform their jobs, and with guests to collect information on how to improve our selections. It’s understanding winemaking practices and their effects on the cost of the final product and deciding what your personal principles are when it comes to purchasing.
Even effective blind (or deductive) tasting relies on a strong set of logic skills. Accessing sensory memory and combining it with a knowledge of regional practices, biological processes, geography, and climate to make a quantitative and qualitative analysis of a wine is not an easy feat without active learning and diligent practice. Being born a sommelier is just as impossible as being born a data scientist — you have to grow into your skills and adapt to the changes as they come.
Just like any iterative process, wine is a web of considerations and logical choices before anything even ends up in a glass. A wine’s acidity is a likely indicator of the climate in which it was grown. Its smell informs of what vessel it was aged in, and using oak is more expensive than using stainless steel. The body of a wine is in direct correlation to its alcohol content due to the viscosity differences in alcohol. Cabernet Franc smells like bell pepper because it contains pyrazine, the same compound that gives bell peppers their aroma. Is this wine’s quality in line with its price point? What is its likelihood to sell? Will having it on the menu create financial room to purchase other, costlier wine due to its profit margin? How likely is someone to drink a light-bodied rosé in the middle of winter when it’s 12°F outside? What trends are impacting the wines that people are buying right now? Is there even room for an Australian Riesling when there is one from New York and one from Germany already on your menu but you just want everyone to know how gosh-darn versatile it is and are hoping that someone who feels the same way happens to pop in for dinner?
Data science speaks to me in many of the same ways that building, curating, and refining a wine menu does in that it requires diving into a body of information, interpreting it in a meaningful way, and presenting it in languages that are comprehensible to people within and outside of your industry. The tools look very different — I never had to use Pandas to open a bottle or to transform my tasting notes into a visualization (although that’s not a terrible idea) –- but in many ways, the approach is very similar. Not only is technical aptitude key, but the ability to transform and communicate information is one of the most important qualities that a good data scientist needs.
My passion for wine opened the door to curiosity about the world and the human effect on it, from large-scale questions such as industrial agriculture’s environmental impact to daily occurrences such as the likelihood of packaging influencing an individual purchase. The need for competent data scientists is rising across industries and strong analytical skills will never be useless — but also like wine, the concept of “data science” is nebulous and terrifying without the right framework with which to start understanding it. My goal is to take advantage of that framework to contribute to global accessibility of the information that we take in every day.
