By Meta Careers
Meet Veronika, who leads analytics for the team at Facebook charged with keeping the community safe and secure. Here Veronika shares her journey to Facebook, what she’s working on and challenges her team faces.
What do you do at Facebook?
I lead analytics for the team that works to build a safe and welcoming community by minimizing bad experiences for people on Facebook. While most people come to Facebook with the intent to connect with friends and family in a respectful manner, some people come with a malicious intent, or don't know that what they’re posting is not allowed on Facebook. They might be involved in activities that violate our Community Standards like creating fake accounts, posting hate speech or trying to scam people. Our team’s goal is to detect and stop these types of activities.
What kind of approach does your team take?
We believe in a formula of Understand, Identify, Execute. First, analytics and research analyze the space we are trying to address. Second, analytics helps identify where we should focus by sizing and prioritizing different opportunities. Third, analytics partners with other functions (engineers, data engineers, product managers, content strategists, etc.) to execute on these opportunities. For example, analytics identifies some of the features that need to be incorporated into the machine learning algorithms built by engineering to help detect bad content or accounts.
Analytics drives the development of experimentation frameworks that enable us to measure the performance of new and existing models. We develop metrics goals, measure progress against them, and identify key contributors to progress. We also work with Facebook’s communications team to create more transparency around the problems we tackle.
How did you end up in this role?
I have been working in analytics and risk management for over a decade. Pursuing a career in analytics enabled me to work in a variety of industries - economic consulting, fintech, online travel, and now building online communities. Over the years I found that I gravitate towards purpose-driven problem spaces. I spent the first year at Facebook building out the analytics team that works on solutions like Charitable Giving and Crisis Response (our tools that help people let friends and loved ones know they are safe in a disaster). As I finished building the team, I was offered an opportunity to grow an analytics team focused on safety and security. I was excited to take this on and work on something that matters so much and has real world impact. It is also an opportunity that enabled me to work on very complex, intellectually fascinating problems.
How does your team stop these bad activities?
We leverage our community of users to report bad material or accounts on Facebook.
We also protect people by analyzing account behavior and posts. For example, photo hash matching technology helps us remove images that have no place on Facebook. As another example, natural language processing (NLP) can help identify text that might be advocating terrorism.
For accounts, we detect certain types of user activity and other account characteristics to assist us in identifying bots or fake accounts. Machine learning (ML) can also aid us in finding accounts that are similar to known bad accounts, such as malicious fake accounts.
What are the challenges you face?
First, it is an adversarial space. The modus operandi of our attackers continues to evolve. Today, they might find a loophole in our Pages product. Tomorrow, it might be a new feature in the Groups product. As we implement new solutions, attackers respond with new behavior. So, our tactics have to quickly evolve as well.
Second, we have a diverse community of users that spans different legal systems and cultural norms. Balancing these is very challenging. In Germany, for example, the law forbids incitement to hatred; you could find yourself the subject of a police raid if you post such content online. In the US, on the other hand, many kinds of speech are legally protected under the U.S. Constitution.
Third, measurement is very hard for these kinds of issues. For a regular product, measurement is a matter of logging. If you log it, you can report on it. But how do you size the prevalence of, say, violent images on your platform? The best detection tools usually can only find a portion. We’ve had to come up with inventive ways of getting a more comprehensive picture of the extent of a problem and how much progress we are making in addressing it.
Expanding on the example of hate speech, what makes it so difficult to eradicate it?
Generally we define hate speech as attacks on people based on protected characteristics like race, religion or gender. Hate speech is one of the more challenging areas because it requires a lot of cultural context in addition to language skills to identify. For example, you can try to proactively detect slurs and take down posts or groups that use them. But that is how you can end up taking down self-referential slurs, like lyrics from a rap song.
Sometimes we see new types of attacks appear. For example, with the immigration crisis in Europe, we started to see posts that dehumanized immigrants. And with the conflict between Russia and the Ukraine, the words the two sides use to refer to each other gained a new meaning, becoming offensive.
Many hate groups use code words in their names and their hate speech. For example, some white supremacy groups use “18” to refer to Adolf Hitler or “444” to refer to "Deutschland den Deutschen” (which translates as “Germany for the Germans”). So, we have to build tools and processes that separate innocuous use of these numbers from hateful use; it’s a constant effort to keep up with the evolution of specific phrases and code words.
What's next for your team?
We are an international team based in London, Menlo Park and Seattle. And we are growing! As Mark Zuckerberg, our CEO, recently said, we're doing a lot in this space by investing in people and technology and we are going to do more.
We already expanded our investment into analytics, engineering, product management and all other functions this year. We plan to grow even more next year. For analytics, we’re investing a lot into the London office and plan to double the size of our team there. We’re also growing in the Menlo Park and Seattle locations. These investments will enable us to go after a broader set of issues and be even more proactive and global-minded about detecting various violations.
Stay Connected
- Learn about Life at Facebook on Instagram via @facebooklife.
- Join our Facebook Careers Talent Community Page for the latest updates.
share-externalRELATED ARTICLES
Careers
Follow us
Equal Employment Opportunity
Meta is proud to be an Equal Employment Opportunity employer. We do not discriminate based upon race, religion, color, national origin, sex (including pregnancy, childbirth, reproductive health decisions, or related medical conditions), sexual orientation, gender identity, gender expression, age, status as a protected veteran, status as an individual with a disability, genetic information, political views or activity, or other applicable legally protected characteristics. You may view our Equal Employment Opportunity notice here.
Meta is committed to providing reasonable support (called accommodations) in our recruiting processes for candidates with disabilities, long term conditions, mental health conditions or sincerely held religious beliefs, or who are neurodivergent or require pregnancy-related support. If you need assistance or an accommodation due to a disability, fill out the Accommodations request form .