Meet Milan S., a research scientist manager supporting Meta’s Statistics and Privacy team, which is using privacy-enhancing technologies and deep learning to build new ML and AI systems that enable consumer experiences that are both personalized and private. We sat down with her to learn about her team, their work and why Meta is the best place to build privacy-focused solutions.
What's unique about your team’s work at Meta?
We focus on tackling emerging needs for our business and developing scientific approaches to prototype scalable solutions. Team members have varied areas of expertise like machine learning, cryptography, statistics, just to name a few, but also strong communication skills to transcend traditional academic and functional boundaries. This combination enables collaboration within our team and with cross-functional partners to quickly solve hard challenges.
While focusing on long-term impact, we’re empowered to tackle a wide spectrum of work. We use that latitude to invest in theoretical research, prototype production systems, and engage with open-source communities. Our team members come together to take on different roles in each project to push innovations from ideas to reality. To get a better sense of what the team does, check out which roles we're hiring for now.
What do you enjoy most about your team at Meta?
The opportunity to explore new methods and develop new products within a culture that balances exploration with responsibility to move fast and drive impact. Our projects can take a long time to materialize—many months or even years—because we're exploring cutting edge technologies and making them practical for real world applications. We are shaping an important direction of the company, and potentially a part of the industry as well.
What exciting problems are you trying to solve?
Traditional machine learning has relied upon the centralization of data, which has led some people to believe that highly personalized experiences and consumer privacy are zero-sum. Our team is investing in the field of privacy-preserving machine learning, developing techniques and building systems that achieve both privacy and utility goals to serve the needs of billions of consumers and more than ten million advertisers. We tackle a range of approaches and technologies—on-device computation, multi-party computation, differential privacy, trusted execution enclaves—because no single approach is going to be the best choice for every challenge.
How has your education and prior experience played a role in what you do now?
My PhD training gave me a strong foundation in mathematics and the experience of tackling hard, previously unsolved problems. I also spent several years at Airbnb as both an individual contributor and as a people manager, tackling novel algorithmic challenges in a two-sided marketplace. It was during those years that I learned how to balance theoretical rigor and product constraints. Since joining Meta, it’s been easier to leverage my expertise in statistics and machine learning to dive into new domain areas.
What do you know now as an employee that might have made you join Meta sooner?
The culture and incentive structures across the company really encourage innovation and collaboration across teams. For example, through our internal Workplace groups, you can easily tap into the latest work being done by other teams across orgs that is interesting to you. Many Meta project collaborations are started organically by team members connected through common interests over a particular application or certain technology, which is extremely helpful for an applied research team like us.
Also, something that’s not typically seen at a large tech company like Meta is the ability to lead in areas of methodology and development without restrictions imposed by tenure, level, or role. Our team’s channel of communication is so transparent and flat, we make it easy for anyone across the company to weigh in to share their methodology expertise.