Six years ago, while pursuing her PhD in management information systems from Purdue University, Na Zhang did a three-month internship with Amazon’s fraud detection machine learning team.
“That was a very rewarding and eye-opening experience, something I will remember for a very long time,” Zhang says.
2021 Amazon Science Internships
Amazon currently offers science internships year-round. Projects will depend on a student’s area of research and interest, as well as the team to which they are being placed. Find out how to apply.
Fast forward six years: Zhang is now an Amazon applied science manager, and leading a team that partners with the company’s product-discovery organization to improve the delivery experience, or options for how quickly a customer can receive their order. Zhang’s team is preparing for two PhD interns this summer from the University of California, Riverside and the University of Texas at Austin, each of whom will be working remotely.
Zhang’s interns are two of more than 8,000 interns Amazon will host (virtually) this year, the largest intern class in the company’s history. More than eight percent of those internships will be for applied research, and data science roles within many of the company’s organizations, from devices and AWS, to consumer and finance.
Amazon offers science internships year-round. Projects depend on a student’s area of research and interest, as well as the team to which they’re assigned. The majority of the company’s science-related internships last between 12 and 16 weeks.
“When I was an intern, I was working with deep neural networks to help identify fraudulent transactions. I got to use state-of-the-art technologies, got an opportunity to experience how industry handles terabytes of data, and how to make systems more scalable.”
Zhang also experienced how her work could have an immediate impact on customers, and discovered her passion for having customer and business impact at scale.
“That’s the part I’m passionate about,” says Zhang. “It’s great to see how your work has impact, and is valued by customers.”
The experience led her to return to the same fraud-detection team a year later. One year after that, the machine-learning model she helped develop as an intern went into production, and two years later she co-authored a paper with colleagues that was presented at the company’s internal machine-learning conference.
Zhang’s advice to the two interns who will be joining her team this summer is to “make the most of your time here, learn as much as you can, and talk to as many people as you can to get the help and support you need to succeed.”
While a virtual format may present some challenges in connecting with other interns, Zhang will encourage the interaction by creating Amazon Chime rooms where her interns can interact, organizing virtual team events, and fostering technical knowledge sharing. Zhang met several other interns during her three-month internship six years ago. “We’re still very good friends now,” she says.
Papers and production
Xin Luna Dong is a principal scientist, leading the team developing Amazon’s Product Knowledge Graph. Similar to the previous two years, Dong’s team will have 10 interns in 2020, nine this summer, and one more in the fall.
But what won’t be similar is how this year’s interns will be working remotely, which will be a different experience for the interns and her team.
“Our team is brainstorming ideas of how to make these internships as interesting, productive and rewarding as they would be if everyone was able to join our team in Seattle,” Dong says.
Dong, who did internships with Microsoft and Bell Labs while getting her PhD in computer science from the University of Washington, is a strong supporter of the program, not only because of how her own career was influenced by her internship experience, but because of how intern projects have helped shape her team’s current work.
Dong cites two initiatives that started as intern projects, but have now grown to sizable programs for her team. “Our interns have done a great job,” Dong explained. “They have planted the seeds, and eventually those seeds have grown to become big projects.”
Last year, Dong and team organized intern projects related to the team’s three “big bet” goals. This year, Dong and team evaluated where their product knowledge graph project will be at year’s end, and pinpointed technical gaps that need to be addressed.
“We identified three big categories, and then outlined intern projects for each category of technical gap,” Dong said. “Our team really appreciates how interns can help us achieve our long-term goals.”
As is standard practice for all interns, students joining Dong’s team get a launch plan on their first day, outlining what’s expected.
“We have two goals for our interns: one is that they publish their work in top-tier conferences, and the other is that their work will eventually make it into production. Our interns who achieve both goals are very excited, and typically are more excited about seeing their work make it into production than publication.”
Recent results have been impressive. Dong reports that eight of ten 2019 intern projects resulted in a published research paper, and five of those projects are moving toward production. This year, she says, her team has published nine papers, and three tutorials have been accepted at top conferences. Interns contributed to each of these projects.
Lunch dates with scientists
Dilek Hakkani-Tür is a senior principal scientist within the Alexa AI organization, and her team focuses on development of natural dialogues with machines. This spring, her team has had two virtual interns, and three more interns will join her team this summer with two more in the fall.
“We love interns,” said Hakkani-Tür, who did an internship at SRI International while getting her PhD in computer science. “Oftentimes they bring a fresh perspective to the problems we’re thinking about, and they bring a lot of energy. They are only with us for a short time, and they want to do well, and we want them to as well.”
Interns working with Hakkani-Tür’s team this summer will be focusing on research related to common sense reasoning for social interactions, and human-robot interactions.
Hakkani-Tür believes interns can benefit from having multiple mentors, so she’s introducing the notion of interns making virtual lunch dates with several scientists within the Alexa AI organization.
“It can be more difficult to interpret feedback from multiple people, but I think it’s beneficial in the end because the students get to balance the feedback from each of these scientists, not just a single individual,” she says. “Those interactions will hopefully help them to make future decisions about their careers.”
Hakkani-Tür’s goal is that each intern will publish at least one paper related to their work. But oftentimes, she says, what really excites them is the ability to see the impact of their science work, which is why she tries to expose them to as much of Amazon’s science culture as possible, so they can see first-hand how teams focus on turning their ideas into production.
How to apply
Amazon’s Graduate Research internship program includes mentorship, moderated discussion groups, opportunities to connect with fellow interns, fireside chats with senior leaders, and a variety of networking events.
If you’re a student with interest in an Amazon internship, you can find further information here, and submit your details for review. Students can also learn more about internship opportunities on the Amazon Student Programs Twitch channel.
