Originally published on Amazon Science.
Eugene Yan’s career path has taken some unusual turns, but his motivation has always been the same: understanding people so he can help them. A policy analyst turned data scientist, Yan is now an applied scientist at Amazon using customer-behavior data to help recommend the best products. In the world of machine learning, however, he’s best known for the way he writes it all down. On his personal site, Yan covers a range of professional and technical topics like machine learning systems, data science methodology, dealing with imposter syndrome, and building data science teams.
Eugene Yan started eugeneyan.com in 2020, focusing on general machine learning and career content. Initially it was for personal development, but then people started reading, and now writing posts takes up the majority of his leisure time. He started the site for personal development, but then people started reading, his network started expanding, and now writing posts takes up the majority of his leisure time. “It snowballed,” he said. “Writing helps me learn better. And when I share it online, it attracts like-minded readers and helps me make new friends. ”
Born in Singapore, Yan studied psychology at Singapore Management University. “I was curious about people, how they perceive, and how they behave,” he said. His college research focused on how competition affects people differently, motivating some and intimidating others. After college, he joined the Singapore government as a policy analyst sifting through legal cases and trade agreements. But it wasn’t long before he began to miss crunching numbers and following the data on human behavior. “I envied my colleagues in commodities who used on numbers and worked with spreadsheets,” Yan said.
He decided to try and make the switch to a data science position based on some familiarity with the subject from his undergraduate research days. He landed a position at IBM in 2013, and from there he moved to data science roles at Lazada, a Southeast Asian e-commerce site, and then UCARE.AI, a healthcare startup.
“In every change in my career, what drives me is helping people,” Yan noted.
At IBM, it was helping people find new roles. At Lazada, it involved helping people find products they need. At UCARE.AI, it entailed predicting chronic diseases and preventing high insurance payouts. “This brings me more satisfaction than dollars and cents,” he said.
While at Lazada, Yan decided he needed more training in the fundamentals and pursued a master’s in computer science from the Georgia Institute of Technology. He graduated in 2019, and then he and his wife began considering a move overseas. He applied for a position at Amazon, drawn to the company’s leadership principles and the ability to help customers read more. He relocated to Seattle to join Amazon in 2020.
While Amazon has several ways to help readers find books, from Amazon Book Review to Amazon Charts, Yan is part of a team developing the recommendation systems that power the widgets behind the Amazon Store’s personalized book suggestions. “Customers tell us what they like based on what they do,” he explained. “They browse for a specific book, a genre or a topic.” His team uses those signals to help surface additional books a reader might like. Ultimately, Yan and his team want to make reading easier.
Early in his transition to data science, Yan started interviewing mentors for advice, some of whom were “rock star data scientists.” He asked what skills he should cultivate to be successful. The one skill a majority of mentors suggested was communication. The people he spoke with emphasized how communication becomes more and more important as you rise in the ranks. “I was like, ‘Are you kidding me?’ But more and more mentors said the same thing,” he recalls. “I thought, ‘This can’t be right, but I’m just going to try it.’”
Yan started practicing his writing, first publishing to a WordPress site. He wrote dozens of posts unnoticed, but then in 2020 created eugeneyan.com and started writing more general machine learning and career content. His writing began to gain an audience. Posts like “Unpopular opinion — Data scientists should be more end-to-end” received more than 500 likes on Twitter. One post on note-taking received 35,000 unique views in a single day. Feedback and praise began to pour in, and his “practice” website swelled into something much bigger.
For a brief period, Yan tried to sustain this level of social engagement. He wrote to please a mass audience and get clicks. “That quickly became unfulfilling,” he said. Now he focuses his writing on topics he wants to learn and aims for an audience of people he’d hope to be friends or colleagues with. “I might have fewer readers now since I’m choosing more technical topics, but these readers comment, disagree, and email me. Each comment and real relationship are worth more than 10,000 likes,” he said.
Yan’s decision to become a better communicator and writer is especially valuable at Amazon. “The writing culture is rigorous at Amazon,” he said.
Amazon’s working backward method starts with an individual or team imagining the product or service is ready to launch. The individual or team’s first step is to draft a press release announcing the product’s availability, and explaining its significance. Moreover, meetings often start with participants reading a six-page document about the meeting’s topic before discussion begins.
Finding your voice and niche doesn’t happen overnight — you have to write and share your work. So just start somewhere, anywhere, and keep writing. — Eugene Yan
“I write as many documents as I code,” Yan said. Recently he received feedback that one of his design documents was easy to understand and clearly laid out everything the reader needed to know. In this way, his writing skills complement his design and machine learning skills. He also started a new site, Applying ML, which includes interviews with machine learning practitioners.
Yan is often asked by aspiring writers for advice on how they can improve their skills. The number one piece of direction he offers is to write for yourself — what do you want to learn and clarify your thinking on? — rather than social engagement. The second piece of advice: “just write.” The best way to figure out your niche and your audience is to simply put fingers to keyboard and start practicing, he said. Maybe after a dozen — or a few dozen — pieces you find your voice, what you want to write about, or what resonates best with the people reading along.
“If you never start writing, how will you know? Just like Blue Origin’s motto ‘Gradatim Ferociter’, which means ‘Step by step, ferociously’. Finding your voice and niche doesn’t happen overnight — you have to write and share your work,” he said. “So just start somewhere, anywhere, and keep writing.”
Join 2,600+ readers getting updates on data science, data/ML systems, and career.
Welcome gift: 5-day email course on How to be an Effective Data Scientist 🚀