It’s hard to imagine there’s ever a slow day at Graphika.
The NYC-based firm has had a hand in exploring how fake news was spread through social media prior to, during and after the 2016 election for the U.S. Senate Select Committee on Intelligence; how censored content from the government in Yemen spread through Twitter; and, in Nigeria, how and why certain hashtags take off — and others don’t.
Built In NYC spoke with Graphika’s VP of technology, a senior software engineer and a science researcher about the real-world problems Graphika is helping to solve, the methods behind sorting through the madness, and the kind of team they’ve built in the greatest city on Earth.
FOUNDED: 2013
WHAT THEY DO: Graphika maps and analyzes the complex fabric of social network structures, or what they call the “cybersocial terrain.” With its patented technology, clients are able to gain an understanding of how information and influence move through networks.
WHERE THEY DO IT: NYC
IN THE NEWS: Graphika’s CEO John Kelly and Chief Innovation Officer Camille François worked with researchers from the University of Oxford to see how Russia’s Internet Research Agency used social media to influence the 2016 election.
A PEEK INTO PERKS: A notable few include an unlimited PTO policy and 100 percent employer-paid healthcare for employees.
Chris Cotter, VP of Technology
Chris is responsible for the overall technical vision and leadership of the growing technology team.
STRATEGO: A collector (and player) of strategy board games, Chris has over 100 in his collection. He’s been focusing on quicker and lighter-lift games as of recent: “It’s hard to spend an entire afternoon or evening playing a single game when you’re a dad of a young kid,” he says.
Talk to us about the problems you and your team solve when you come into work and the technologies you're using or building to accomplish those goals.
We’re taking technology coming out of Graphika Labs, which is doing grant-driven research, and turning it into products usable by everyday people. At the same time, we need to make sure we’re not losing the scientific rigor that gives our clients trust in what we’re building.
One exciting thing about working here is actually seeing the impact we’re having in the real world. I get to see both the public acknowledgment of our efforts online and hear from our customers of additional impacts they’re able to produce that don’t always come out in public.
Tell us about the breakdown of your Innovation Team and how the sub-teams work together to carry out Graphika’s mission.
Graphika has an Innovation Team that consists of a Labs Team on one side and an Analysis Team on the other. Both of these teams are pushing forward the cutting edge of what is possible in online network analysis, and it’s the product and tech team’s responsibility to integrate that into our core platform when it’s ready. The work from the Labs Team is generally grant-driven research that is done in collaboration with other research teams around the world. The Analysis Team is working with organizations and brands to find useful information and derive new analytic techniques against it.
I get to see both the public acknowledgment of our efforts online and hear from our customers of additional impacts they’re able to produce that don’t always come out in public.”
What’s a major challenge facing social networks, particularly disinformation? How are you and your team trying to help overcome it?
Social networks operate at a scale that is past the average person’s ability to cope with. We have learned behaviors around trust and social proof that help us navigate real-world social situations. When you take these behaviors and apply them to online communities, they break down in fundamental ways. Our platform helps to overcome this by breaking down the discourse around topics into their fundamental components at scale and letting people understand how those components interact in comprehensible ways.
Bridget Keyes, Senior Software Engineer
As a back end engineer, Bridget writes code for Graphika’s data services and API.
FINDING ZEN: Nearly every morning Bridget does yoga, which she says is a calm way to start her day.
What makes working at Graphika — as well as your day to day — so unique?
I’ve worked at a lot of different places, including a small nonprofit, a large defense contractor and a hedge fund. Graphika is a unique environment because I get to interact with researchers, analysts and people from different backgrounds, not just programmers. We work on interesting problems with interesting people. The pace is pretty fast here and we get a lot done with only a few people — plus, we get that chance to really own our code and the decisions around it.
How is your team working to evolve Graphika’s products? Are teams cross-collaborating to make this happen?
The technology team is responsible for building and maintaining our products. Recently, we have been working on an improved search feature, and our front end, back end and product teams have been working closely with our customers and internal analysts.
We also often work together with analysts to do custom data analysis and visualizations. The research team is comprised of scientists building cutting-edge scientific tools, and the tech team works with them to help them build and scale their ideas.
Graphika is a unique environment because I get to interact with researchers, analysts and people from different backgrounds, not just programmers.”
Where do you see your company in a year’s time?
I think Graphika will look quite different in one year. We’re adding an all-new front end, so it will literally look different. Also, we’re always working on adding new data inputs and discovering new insights. Of course, we’re also improving the code so we can handle all our data more efficiently. There’s a lot of interesting challenges to work on!
Amruta Deshpande, Science Researcher
As a member of the Graphika Labs team, Amruta’s day to day involves conducting theoretical and technical investigations into Graphika’s research objectives, and she also maintains the code base of a product currently in production.
HAVING A LAUGH: Comedy is all about ... timing. Amruta loves breaking down comedies and stand-up sets to understand the timing and structure of a joke, which bleeds into her being more observant in her daily work.
You work on the Graphika Labs team. What’s your day to day like, and what kinds of projects are you currently working on?
I do a variety of tasks on any given day. My favorite is discussing problems and solution strategies with my manager, Vlad, the tech team and other colleagues. I do some project management for a part of one of our research collaborations. I code to maintain one of our products in production. And other tasks I do, that are more typical of a researcher position, include reading literature, learning methods, learning software tools and, finally, performing data analysis and modeling.
Some of our current projects are in developing causal inference tools on social networks, developing general tools for network structure analysis for applications to our main product, as well as for studying disinformation strategies, coordinated information spread and contagion monitoring.
My favorite [daily task] is discussing problems and solution strategies with my manager, Vlad, the tech team and other colleagues.”
Explain how your team uses computational social science to dissect social networks. What role does this play in your work?
We use a combination of analyses that are well known in the fields of computational social science, network science and data science, along with some custom metrics developed in-house to dissect and study social networks. Developing custom metrics has been key to our analysis efforts here at Graphika. When we address a question for which the cybersocial terrain is well defined, then we often start with our core product, a map, which allows us to do context-based analysis and gives us the added advantage of using our pre-computed metrics. When the question we address is more theoretical — for example, from a grant-related project on simulated data — we use existing methodologies or build models as needed.
Tell us about a project or challenge you’re working on that excites you most. How are you solving that challenge?
I’m most excited about causal inference being integrated into our existing tool set. Causal inference is a strictly mathematical formalism, and experts in this field are often called upon to give real bounds on understanding causes of observed outcomes. Scientists have only recently been working on applying this to social data and have shown that it can be used to find out why something in the social network is the way it is. There are a great number of challenges that underlie this marriage of causal inference and computational social science.
We are collaborating with experts at Johns Hopkins University, Whiting, and others at Oxford who are helping us to explore integrating causal inference methods into our current toolset. We hope to identify some clear cases where these methods will be regularly applicable.
