No PhD, no problem: One software engineer’s path to applied science
Sneha Rajana is an applied scientist at Amazon today, but she didn't start out that way. Learn how she made the switch, and the advice she has for others considering a similar change.
[Editor's note: In March 2022, Sneha Rajana accepted a role with Meta.]
In the ever-changing tech sector, jobs are constantly evolving to match new markets and skill requirements. That means career transitions are increasingly common, but no less daunting. Sneha Rajana could teach a master class on the subject: In just three years, she went from being a newly hired software development engineer (SDE) to a position as an applied scientist at Amazon. And, she said, laughing at the most common assumption about taking on that role: “No, you don’t need a PhD.”
Rajana graduated from PES University in Bangalore, India, in 2015 with a degree in information and science engineering, and then went straight to the University of Pennsylvania School of Engineering and Applied Sciences for graduate school.
“During my time at Penn, I specialized in machine learning and natural language processing, and also did a graduate thesis. So that was my first research experience in this space,” Rajana said. She interned at Audible in 2016 as a software development intern and graduated in 2017 with a degree in computer science and engineering, specializing in machine learning and natural language processing (NLP).
“My graduate thesis went really well,” she said. “I also published two papers at two very well-known NLP conferences [the Association for Computational Linguistics and the Conference on Empirical Methods in Natural Language Processing] at the time.”
Getting started at Amazon
When Rajana started her first job at Amazon as an SDE, she worked on an outbound marketing team. “I didn't really know what an applied scientist was, or you know — what those roles were,” she recalled.
For a year, her work drew upon her engineering skills. “I learned so much about scalability, and how to write code to create products for billions of customers,” she said. “I love engineering, and I love to code and write code in production.”
In 2018, she moved into personalization for visual categories like furniture and fashion, where customers have a lot of preferences. That team worked to capture those desires to show them relevant suggestions in real time.
“I worked on a lot of recommendation models — the engineering side of things, a lot of data stuff,” Rajana said. “And I was exposed to applied scientists in this new team. My senior manager, Soo-Min Pantel, was an applied science manager.”
That’s when she learned that maybe she didn’t have to choose between research and engineering — she could have a role involving both. She hadn’t forgotten her love of machine learning research and thought she might be interested in an applied science position.
Mentors are key
“I started talking to a lot of mentors, a lot of applied scientists across the company,” Rajana said. She also taught courses as part of Amazon’s Machine Learning University, including Natural Language Understanding, Introduction to Data Science, Introduction to Reinforcement Learning, and Feedforward Neural Networks.
“That’s when I really built up my network of people that I knew who were doing science at Amazon,” she said. “I set up a lot of chats, and that's when I learned that there were different roles.”
Rajana found these mentors to be invaluable in helping shape her path forward. Their experience in their own careers meant they were uniquely positioned to give her advice and offer perspective. Mentors also provided a low-stakes sounding board.
“My mentor, Sebastiano Merlinos, was a senior software engineer at the time, and I brought transitioning to research up with him first, before I brought it up with my manager,” Rajana said.
Merlinos, who was both working toward becoming a principal engineer (his current role) and had collaborated across a number of teams, was able to introduce her to others. “He connected me with at least three or four scientists from his previous teams, or other people that he knew.”
As she spoke with these scientists, she realized most of them had PhDs. But they told her that while a PhD gives you some skills, it’s not a requirement — and that she brought other skills to the table as an engineer.
“I know how to take an idea, right from prototyping to production,” Rajana said. “You can work on the most complex machine learning models, but if you can’t actually bring them in front of your customers, be able to put them in production, and be able to scale for billions of customers, then that doesn't really work, especially at a customer-obsessed company like Amazon.”
Through her mentor network, she found others who had followed similar paths from software development to applied science — without a PhD. “So, while it was hard, it didn't seem impossible to me,” she said.
A little help from her manager
Buoyed by the fact that there was a path forward, albeit a less-traveled one, Rajana determined it was time to speak with her manager, Pranav Varia, a software development manager. “I was working in a team where all of us had similar skill sets and were working on similar problems. So I had a talk with my manager about how to pursue a horizontal transition,” Rajana said.
See our latest science career opportunities
We're looking for research scientists, applied scientists, data scientists, economists, and more. Find out where we're hiring.
She then worked on some projects that could demonstrate her scientific chops, publishing in the Amazon Machine Learning Conference. She also submitted a few patents for her work. “Those things really gave me more confidence and showed that I would succeed as a scientist as well,” she said.
Rajana then waited for the right project to come along. When it did, instead of bringing in an applied scientist, Varia asked Rajana if she would like to explore whether that was the kind of work she really wanted to do before she made the official switch.
“I worked on a product recommendation strategy by building a technology that can detect implicit shopping intent for a customer during their shopping journey by automatically identifying key attribute preferences, and delivering more relevant product recommendations. This recommendation strategy adapts in real time as customers’ key preferences change as they browse more products,” she said. “It was a very interesting problem.”
The advice I would give to others is to start the conversations much earlier than I did.
The career-transition road wasn’t without it bumps: Rajana says that there were times she felt uncertain, but that reaching out for support was key to moving forward. When she needed to brush up on some topics as she transitioned to her new role, she took courses, refamiliarizing herself with some areas she hadn’t studied since college — but she didn’t do it alone. Again, her mentors came in handy.
“I had many different types of mentors, some long-term mentors who helped me overall with the switch and who continue to help me,” she said. “But I also had some shorter-term mentors who helped me specifically with the problem on which I was focused. It was really helpful to have built up that network.”
Her perseverance paid off when in May 2020 she assumed her new role as applied scientist. “Since then, it's been really exciting,” she says.
Rajana said others who are interested in making a similar change should consider moving faster than she did: “The advice I would give to others is to start the conversations much earlier than I did. I think I definitely spent more time than necessary just debating whether or not I should even have that conversation.”
And while she draws lessons from her past, she remains excited by the future. In 2021, Rajana will be focused on creating new visual front-end experiences and backend, science-based algorithms for recommending relevant products to Amazon customers.