Meet Rhea Goel, Senior Applied Scientist in Sponsored Products

Meet Rhea, Senior Applied Scientist at Amazon Ads. She started her career journey as a software development intern and now manages a team of scientists and engineers.

In this interview, Rhea discusses how she has combined her engineering background and machine learning expertise to tackle some of the company's most complex advertising challenges. She also shares how scientists can shape business strategy while growing their careers in unexpected directions.

I started as an engineering intern. I always had plans to transition into a science role because that's what my education was in. However, I purposely chose to start as an engineer so I could learn the skills to build and deploy production systems and understand the best practices to build for the real world.

After transitioning into an applied science role at Amazon Fashion, I specialized in recommender systems, ranking, and personalization. Now I'm pursuing a path to management in the science family. Being a strong engineer has helped me become more autonomous and confident as a scientist, and it also means I’m very capable of leading a diverse business team that has both science and engineering specialists working closely together to achieve goals.

In advertising, there are really hard problems to solve. You're trying to place ads in a way that ensures that customers see the most useful content for them—while at the same time advertisers get the most return out of their ads and Amazon, as the publisher, grows its business. Balancing between three entities is very challenging and nuanced. I particularly enjoy the fast experimentation culture at Amazon, where we get to test new machine learning models that operate under these diverse, often competing, objectives.

At Amazon, applied scientists are really able to influence wider business strategy. There are several formal opportunities during the year, including annual planning, quarterly planning, and regularly scheduled hackathons, where anybody from junior scientists to principals can bring ideas to the table, working backwards from a customer pain point or business goal. At every level, we’re encouraged to put our ideas to paper for leadership to review and take forward.

There are so many practical applications of large language models (LLMs) that could transform the ad ranking space. For example, we're currently exploring how to use LLMs to better understand shoppers’ search queries to produce more relevant search results in terms of different product attributes such as brand. An LLM simplifies this for us because it comes with a lot of knowledge of the world out of the box.

I feel that leadership is very invested in my career growth. For example, a while back my manager mentioned to my director that I was interested in pursuing the management path. My director remembered this, assessed my skills over time, and then brought me an opportunity to make the leap. I’ve recently started managing a team, which has been a great learning experience.

There are also plenty of mentoring opportunities. As soon as you join Amazon, your manager typically assigns you an onboarding buddy and a mentor. There's a Women in Engineering mentor program and formal Amazon-wide mentoring schemes; your manager can help facilitate finding the right mentor for you.

Absolutely. Every year Amazon holds the Amazon Machine Learning Conference (AMLC), which is an internal science conference with a very high bar and low acceptance rate. Scientists often work on projects with intellectual property and it may be hard to publish externally, but with AMLC, you get a chance to publish scientific research. Given that it has such a high bar and low acceptance rate, it’s just as rewarding. If you get selected for an oral presentation or a poster presentation, you get to present your work across Amazon, which is great for visibility and personal growth.

I recently worked on a model that customizes the Sponsored Products ads on the search page to show customers more products that match their interests. Our ethos at Amazon is “Customer Obsession” so our ultimate aim is to make the advertising experience helpful and relevant to customers. Our model improved the customer experience significantly.

This model uses reinforcement learning, one of the harder disciplines to productionize. We had great discussions internally about the overlap between reinforcement learning discipline and causal machine learning. People on the team borrowed ideas from both fields to build this model. It was a rewarding project because it was scientifically challenging and also had a tangible customer impact.

This is probably the most fast-paced environment you'll work in, and it's a place where you can find your own space. If you're interested in a more deep research–based career, there are teams at Amazon Ads that do that. If you're more interested in fast experimentation and business application of the latest machine learning technologies, there are plenty of teams that do that too.

Finally, the Amazon scale is massive, so you get to learn from some of the best minds in the industry. If you're open to learning, you can really become an expert in the field because you're surrounded by the best. Whatever you want to do is available here.