Justin Fine joined Katana Graph in September of 2021 bringing experience in big data gained at Microsoft, Neo4j, and Accenture. Justin’s background in applied math includes a lot of graph and matrix algebra for solving graph-like problems. He later worked in anti-money laundering and fraud analysis in government, banking, and telecommunications. Justin recently spoke with us as part of our employee spotlight series.
Before working at Katana Graph, what was the most unusual or interesting job you’ve ever had?
Justin Fine: I worked in National Intelligence, specifically working with supercomputing and connected networks. Back then it was mostly using MapReduce, Python, and a bit of SAS to write predictive analytics. We were seriously limited by compute capability, and because the size of the data was so massive, Katana is a breath of fresh air.
Is this the kind of career you anticipated having when you were a kid?
JF: There were a few careers that were glorified when I was younger. I’ve always wanted to be kind of an academic, and be a mathematician in that setting. I’ve always been very passionate about the subject, mostly because I didn’t know there was anything outside the five core subjects that you usually learn in school. But that aside, I very much enjoy it. As far as finding some sort of career choice early on, I also liked the idea of problem-solving — things that you don’t necessarily have an answer to from the get-go. So I looked into medicine a little bit until I realized I didn’t like blood. It was kind of a bummer. But I did like studying, and I had some pretty solid teachers that provoked me to study on my own. When you find a subject that you like to study on your own, that usually leads to a good outcome. If you’re doing it in your free time, doing it when you don’t have to, you can just keep going.
You mentioned compute limitations that you encountered earlier in your career. Is that the sort of customer pain point that Katana Graph addresses most?
JF: It is certainly one of the main ones. Scalability centered around the speed of graph algorithms is really key. I also think the ability to easily run machine learning and AI on a graph is a huge differentiator and will come to be even more important in the next few years.
Can you say more about how Katana is able to run machine learning and AI when others cannot, or how is it run differently?
JF: One way it’s run differently is just purely based on the fact that you can separate the compute and storage. Being able to run something that has tens of thousands — or hundreds of thousands — of micro computations that need to happen, that usually is a difficult thing to do because after breaking that up into the components there’s still a “push back together” phase. So I still have to get all of the answers and then push them back together. A lot of the time that was done with something like MapReduce or other technologies where that pulling back is a difficult thing to do; it’s very time-consuming. I would say that a lot of the other graph vendors, specifically those that are either graph-native or have a graph solution on their database itself, a lot of them have an issue with that, right? So the parallelization of tasks and giving a very complex answer back quickly tends to be something they struggle with. Traditionally, the answer has been to throw more money at it. But that only gets you so far, and sometimes the situation doesn’t even allow it.
There are certain SQL solutions that can use a graph data structure, with a graph layer on top of a document store; that can work, but it’s not optimal to say the least. When you get into the terabyte and petabyte size, when you get into the tens of billions, hundreds of billions of nodes and relationships, which are the components in a graph, it becomes difficult. You get different problems, so no matter how many dollars you want to feed into a system like Amazon Web Services, you simply cannot get the speed or performance you need. So we’re offering a way for you to distribute that, and to get those answers back quickly.
What advice do you have for prospective candidates and recent new hires?
JF: Be passionate about what you do and show it. There are a lot of people here with a smile on their face for a reason. For new hires, I’d say don't be afraid to fail, but also don't be afraid to ask for help. People learn in different ways, and we can support many paths to success.
What is your favorite part about working for Katana Graph?
JF: Oh, it’s the people without a doubt. They come with energy and passion about what they do. For me that is huge and it pushes me to learn, help, and grow.
Do you have a motto or personal mantra?
JF: Try to keep a positive mental attitude. It really helps to wake up and think "I'm going to have a great day!"
What was your favorite book, toy, or outfit as a child?
JF: I'm going to cheat a bit and say my favorite authors were Jules Verne and Alexandre Dumas. I was a bit of a shy kid and the escapist element of those types of books allowed me to enjoy a solid protagonist.
Favorite travel spot or best vacation you’ve been on?
The south of France was a pretty wonderful place, but really anywhere around the Mediterranean is a great place to be for food, activities, and weather.
Is it just a coincidence that your two favorite authors also happen to be French? Or do you have a particular interest in that part of the world?
I do love French culture. Those authors are escapist authors, which I think was prevalent in that time, in that area. But, no, I’ve found the South of France just very relaxing and I like Italy as well, so I mean that whole Mediterranean coast, I would say, from Portugal all the way over, is amazing. And Monaco is amazing if I had a few dozen more zeros behind my name… I just loved the climate; the people seem very laid back. That’s why I’m in SoCal now, enjoying my work with Katana Graph.