[ad_1]
Editor’s note: Alex Guazzelli is a director of machine learning in Amazon’s Customer Trust and Partner Support unit. Here he shares his thoughts about the essential qualities of a great scientist.
Not long ago, I was in a meeting with a group of science mentors at Amazon. We were sharing ideas on how to make the best of the mentor/mentee experience. A question came up that I thought was worthy of reflection and sharing: “What makes a great scientist?”
During the meeting, my first impulse was to answer it in terms of the Amazon’s Leadership Principles (LPs), a group of principles those of us who work at Amazon live by. In my view, a great scientist is someone who excels in “Learn and Be Curious,” “Invent and Simplify,” and “Deliver Results.”
I have been around many scientists throughout my career and have the pleasure, honor, and responsibility of leading a science team at Amazon in Buyer Risk Prevention (BRP). My team, Payment Risk, is responsible for protecting Amazon and our customers from payment-related fraud.
From the days I built machine learning (ML) models myself, to managing teams that are working on the leading-edge of ML, I came to realize that great scientists are the ones that spend time learning and improving themselves. They do not do that because they want to be promoted to the next level, or simply to learn a new technique to solve a particular problem. They do that because they are inherently curious about the world around them. A great scientist never stops asking “why” and is excited about finding ways to solve a particular problem.
Great scientists are invention machines, they find innovative ways to tackle challenges. In their quest, they are not afraid to fail as that is part of experimentation. However, I see many good scientists approaching problems from the wrong perspective. They concentrate on the problem and lose sight of the solution.
Great scientists are those who constantly try to find the simplest and most efficient solution.
For example, I have seen good scientists get excited about novel techniques and try to retro-fit those to problems, rather than working backwards from the problem and trying to identify the best solution (given constraints). The result: a very complex and elaborate model. Not that complexity is wrong. I have seen complex processes solve big and complex problems, but I have also seen simple solutions do the same.
The quest for simplification, in my view, is also an inherent quality of what makes a great scientist. These days, whenever I am attending a presentation or a talk, the first question that comes to mind relates to Occam’s Razor: Is the approach being proposed the simplest solution to this problem (given various constraints, e.g., performance metrics, business fit, implementation difficulty)?
Great scientists are those who constantly try to find the simplest and most efficient solution.
Moreover, experimentation for the sake of experimentation advances knowledge, but is not that useful if it doesn’t, at the end, benefit our customers. At Amazon, research work is all about delivering results that delight our customers. In ML, it is common to see great experimental results, but without a clear path to production. This issue comes, oftentimes, in tandem with complex solutions that require engineering acrobatics to make a difference, are hard to implement, and difficult to maintain.
Obviously, great scientists also adhere to standards that deliver high-quality solutions. However, that needs to be balanced against a need to continually give customers a better experience.
A great scientist is .. unafraid of competing with others in a healthy way. This involves not only empathy, but also the recognition, attribution, and promotion of the work produced by teammates or scientists in another team.
As I pondered more about what makes a great scientist, I started to think of other dimensions beyond our current LPs. For example, scientists don’t operate in isolation. A great scientist needs to work well as part of a team. Being able to work with others in an effective way requires “emotional IQ” and our differences make us stronger. Recognizing that we all have our own way of perceiving and interacting with the world is an important step toward realizing that others may not see the world as we do. From this perspective, a great scientist is also someone who attentively listens for others’ ideas and inputs. They then use that to confirm, disprove, or change and augment their proposals/solutions.
A great scientist is one who is open to new ideas and interpretations. They can also disagree with others in a respectful and productive way, while engaging empathetically. Empathy is not the same as simply agreeing with someone else. It works by putting oneself in someone else’s shoes with the objective of looking at the problem/solution from another person’s perspective.
Lastly, a great scientist is one who is unafraid of competing with others in a healthy way. This involves not only empathy, but also the recognition, attribution, and promotion of the work produced by teammates or scientists in another team.
If you would like to explore opportunities within my organization, visit our job listings. If you would like to explore all the open science roles within Amazon, I encourage you to visit the Amazon Science careers page.
[ad_2]
Source link
