Welcome to the 2019 Predictive Analytics World (PAW) Conference Series. Recorded live in Las Vegas, this series is bringing interviews straight to you from exhibitors and speakers at this year’s event. In this interview, host Jamin Brazil interviews Ryohei Fujimaki, Founder and CEO of dotData.
Find Ryohei Online:
My guest today is Ryohei with dotData. Actually, you guys have the premier booth location on site at this year’s Predictive Analytics Conference.
Yeah, what do you guys think about the show so far?
Yeah, this show… This is the first time to sponsor this show. And the show is great: like a lot of relevant audiences. And we have the keynote presentation for 20 minutes. It’s very well received. A lot of data scientists are looking for new trend in this industry. So we really like it.
Yeah, it’s pretty well attended. I really like the layout of it. I feel like there’s a nice cross-pollination between the speaking events that are happening and then the exhibit floor. It’s really well laid out. And the attendee list is pretty good.
Yeah, that’s true, that’s true. It’s a very good mixture of the technical audience and the business audience. So, we really have the good conversation with a lot about 10 days in our booth.
Yeah, that’s great. And your booth, by the way, I think is spectacular. It’s the perfect booth kind of as the entry point ‘cause it’s very welcoming and you feel like you just go sit down and have a nice conversation.
And also, we have the Lunch and Learn Session to talk more about data science automation, particularly for Python users. There are a lot of people standing, taking a lot of memos, and they are taking maybe thousands of pictures. So that’s a very, very good session we had.
Oh, that’s great. I’m sorry… I’ve been doing podcasts straight through like every 20 minutes or so. So I haven’t been able to attend any of the content, unfortunately, which is very disappointing ‘cause it seems like it’d be interesting. So, dotData, what do you guys do?
Yeah, so, dotData we are offering end-to-end data science automation. And, basically, we are the first and only company who can automate end-to-end data science process from raw data through data on the feature engineering and machine learning in production. In particular, dotData we have the very strong artificial intelligence technology that automates the feature engineering process. The feature engineering process was told it’s not possible to automate because that’s a black art of domain expert. But we invented the really strong technology, our world is going to explore a lot of business hypotheses with automated expertise.
So congratulations. That is a very hard problem to solve. Do you have a favorite customer story?
Yeah. So, the most favorite story I have is our first customer, of course. That was a very, very exciting moment. Actually, that is the project we kind of decided to launch the company. And that was one of the top 15 banks in the world, a very, very huge bank. And they have the data science team. And their problem is, first of all, they have no sufficient data scientists. It takes a very long time to complete a data science project: each project takes three to four months by a couple of data scientists. What we have achieved in that project was literally we just took out tons of huge table, huge data in the bank, and they applied dotData technology. Just within a day, we delivered the outcomes to the customer. And the outcomes are even better than the result of the data science team. The customer was so impressed and so excited because that is really accelerating their process. It used to take months, but now they can complete a project in a couple of days. It’s a huge acceleration
Even with a better outcome.
Yeah, even a better outcome.
When you’re interacting with a customer, are you interacting with the data science team pretty heavily? Do they see you as a partner? Or do they see you a little bit as a competitor?
Uh, no, we are actually quite good partner because we are always telling them automation is not something to replace a data scientist, but it helps data science team in a couple ways. First thing is acceleration. Just imagine they can run a data science project within a couple of days. Eventually, data science project is turnaround. They have to try a lot of different ideas and learning what works and what doesn’t work. That turnaround agility is a key for succeeding in data science project. So that is one way we are going to help. Another way is what we call democratize data science. Experienced data scientists should very focus on high-impact project or technically challenging project. But there are a lot of templated projects, standard, common projects that even non-data scientists can execute with our data science automation. So data scientists are not our competitors at all; rather, we are helping them be more efficient, more effective.
So that’s your user. I like your framework of this democratization. It feels to me like more and more people are leveraging AI-based technology in order to make informed business decisions. Your tool, obviously, is a great example of this. What other trends have you seen, having been in the industry for over a decade, what have you seen that’s evolved and where do you think the industry is going in the next three to five years?
So, actually, for these couple of years to maybe two to three years, first of all: automation of machine learning and data science is going to be a very, very big momentum because there’s a lack of data scientists, a lot of data science projects that eventually fail maybe because of data scientists, because domain expertise, because communication between business and data science. There’s a lot of reasons. One the other hand, automation can address a lot of these issues. It’s not replacing data scientists, again, while it is going to address a lot of industry challenges. This is the first step. The second step we are seeing just imagine automation enables us to build a lot of machine-learning models very, very easily. Today, we are talking about 10 models, 50 models but two to three years later, we are talking about 100 or even 1000 machine-learning models. What happens is operationalize this model, maintenance of this model: those are going to be a very big problem in the next three to five years. So this is another area we are working very hard.
So the issue there is it’s hard to maintain all these disparate, niche models.
Yeah, because the value to build a model is getting lower and lower.
That’s very interesting. Haven’t actually heard anybody articulate that point before but that’s fascinating.
Yeah, for sure. If somebody wants to get in contact with you or sales at dotData, how would they do that?
Yeah, please visit us at dotData.com or LinkedIn and please download our White Paper or look at the webinar and let us discuss in detail.
It’s been an honor having you on the Happy Market Research Podcast today.
Yeah, thank you very much.
And for all of you who are listening, please take the time to screen shot and share this content. There’s a ton of value wrapped up into these types of conversations as you get just a microview of a major player inside of the analytics space. I hope you enjoyed it, found value. Have a wonderful rest of your day.