|
0xdata |
|
|
February 3, 2014 Issue |
||
|
The Most Powerful Name In Corporate News and Information |
||
|
Big Data Predictive Analytics and Machine Learning In-Memory Platform |
||
|
0xdata develops
open source H2O, the world's fastest in-memory platform for machine learning
and predictive analytics on big data. Running advanced algorithms such as
GBM, GLM, PCA and RF, among others, users can get to interim and final
results quickly to help them make better data-driven decisions faster.
0xdata is based in Silicon Valley and is backed by Nexus Venture Partners
along with other leading angel investors in big data.
CEO,
Co-Founder Sri is co-founder and CEO of 0xdata (@hexadata), the builders of H2O. H2O democratizes bigdata science and makes hadoop do math for better predictions. Before 0xdata, Sri spent time scaling R over bigdata with researchers at Purdue and Stanford. Prior to that Sri co-founded Platfora and was the Director of Engineering at DataStax. Before that Sri was Partner & Performance engineer at java multi-core startup, Azul Systems, tinkering with the entire ecosystem of enterprise apps at scale. Before that Sri was at sabbatical pursuing Theoretical Neuroscience at Berkeley. Prior to that Sri worked on nosql trie based index for semistructured data at in-memory index startup RightOrder.
Sri is known
for his knack for envisioning killer apps in fast evolving spaces and
assembling stellar teams towards productizing that vision. A regular speaker
in the BigData, NoSQL and Java circuit, Sri leaves trail @srisatish. |
|
|
|
Interview conducted by: Lynn Fosse, Senior Editor,
CEOCFO Magazine, Published - February 3, 2014 CEOCFO: Mr. Ambati, what is the concept at 0xData? Mr. Ambati: The company is based on the premise that algorithms are essential to the next generation of applications and platforms. Big data has many good things going for it. Historically though, people have had to choose between modeling or using big data. What we have done is taken cutting edge machine learning and math algorithms to scale on large data sets, irrespective of the number of columns or rows of data. H2O works on a massive scale. So that customers can, for the first time, run their algorithms on large data sets without having to sample and loose the outliers in their data.
CEOCFO: Are most people who are using big data aware of the shortcomings? Are they looking actively for a better solution? Mr. Ambati: We have tremendous attention from the marketplace. It is mostly driven by the pain points of the customer. Customers in the big data space are using very simple algorithms such as count or averages to model their data sets. When they need to use complex algorithms they had to end up using R, open source R or close source packaged solutions. With text data they are able to combine their expertise and knowledge of R, and allow that to run on large data sets at a very, very high scale math engine called H2O.
CEOCFO: What were the challenges in putting together your offering? Mr. Ambati: The thing about math is that math always gives you a number. So accuracy is key. Math also has a high barrier to entry especially in our culture. The second one is distributed in-memory systems are hard. Counting for distributed systems is harder. And re-wiring the algorithms to work across many machines is very difficult. The other big technical problem here is that most of the math has been historically used on small data on a single machine. Therefore, what we did was take the lessons of the 1960s and 1970s, when we did not have that much machine power or that much memory, and bring those to bear on modern machines. It is like we almost went back in time to bring mathematical and engineering novelty to the modern world.
CEOCFO: Are there particular types of companies or industries that would be more likely to use your product? Mr. Ambati: We are seeing tremendous traction from the insurance industry. We are also seeing a lot of use in the retail and online enterprises. Some others are using us for pricing. We are also seeing interest in the oil and gas industry.
CEOCFO: How do businesses find out about you? How do you reach potential customers? Mr. Ambati: Word of mouth. We are focused heavily in building a grass roots movement. We are building a new space here, surrounding the space of big data science. We are teaching much of the audience how to use sophisticated algorithms to model their data. Data is hot. Data is also very messy, very sparse, very unbalanced; it is a “needle in a haystack.” We are educating the business audience along with the product we are building. Therefore, from the very early days of the product we have been evangelizing the space and building the grass roots movement involving all three spaces of math, mathematical physics, (physical interpretation of the math) and software. We have had tremendous grass roots attention, which has led to our adoption by the sophisticated data scientists. Those early customer have been referring us to new customers. Open source has been a great aspect of our go to market. We are the only open source company in our space. We are working very closely and innovating with the Hadoop and the R movements. It has been a great draw for our customers.
CEOCFO: Do you find that people are skeptical at first? Mr. Ambati: The difficult thing is with the big power users. The immediate denials will come from, “No, big data is not that important. We get a lot of our stuff done with ten thousand rows or one hundred thousand rows. We can model it.” The second skepticism is, “I do not need sophisticated algorithms. We get a lot of stuff done without this.” Therefore, what we have done is taken the algorithm to apply that on big data on millions of rows, hundreds of millions of rows, billions of rows. We finish it in five to seven seconds on our platform. That basically means that they are able to run much larger data sets (faster than on smaller data in other systems.) Therefore, when they see the actual product in action and results are ten to twenty percent better predictive power, they immediately convert. There is a lot of skepticism before they actually see the product in action, because there have been many players in this space which promised and never delivered in the BI space or analytics space. They have been promised a lot and never actually saw the products in action. Therefore, when we actually go in and present them something or tell them something, we try not to put too much PowerPoint in front of them. We just show them the product in action on 16-machines in cloud or four regular Intel machines on premise, so they get a very simple deployment model that they can quickly run and see for themselves. Some of our customers download the product and play with the product already before they make the first phone call to us. It has been a product driven, proof point driven and vertical driven customer acquisition story so far.
CEOCFO: How long has the product been available? Mr. Ambati: Our first drop has been March 2013. The company itself was started on March 2012. We started building the core product and then prototypes somewhere last summer. We delivered the first product nine months later and got it into customers and our early users’ hands, so they could give it much more real life. Over the last quarter we have had quite a few more customers than we had anticipated. We are heading into the place where we will be able to really talk about these customers and their use cases in a very bold fashion.
CEOCFO: What surprised you as you have gone through the process of creating the product and commercializing? Mr. Ambati: There were a lot of surprises. I think startup life is full of surprises from day one and every day, I think there are four or more surprises. One of the interesting things that surprised us was that most people are actually excited about simple things, as in very easy math. They are looking at the summary of their data. Most of my customers want a summary of their hundred gigabytes of data that runs on many machines. Some of my customers want to just know what are the principle components using very simple math. The second interesting surprise that we found, which we actually approached to solve is that the variety of data; it is very messy. Simple stuff that you assume would be easy actually turns out to be a much longer drawn out effort. More interesting was the fact that better math does work. Unbalanced data sets are actually very hard to deal with. However, sophisticated new math has emerged in the last four or five years of research. Therefore, when we started with this kind of math, we are seeing that math is actually able to predict far more than what we anticipated. We can actually predict quiet well what movies you are more likely to watch next and what purchase you are likely to make in the next few months. The algorithms are able to figure that out. A third interesting surprise we found was that building one model for the entire user space does not make sense. You want to build models for users separately and build a model for every one of your users. That is a very interesting aspect. Finally, if you really see the kind of adoption that we are seeing, the kind of encouragement we are getting from the marketplace, from our users, from an early community; our community did not expect our confluence to walk through the door. We did not expect our earliest community members to continue to come again and again but now have people that come to seven or ten of our meet ups, out of the many that we organized. You see a “cult” forming around cool. People want to be part of a high impact movement. That is a very, very pleasant surprise - making it easy enough to build a movement in the spirit of a social and highly networked world. As you keep doing things, things keep opening up and people keep opening up. We did not expect this to happen, but it is happening through chance and word of mouth and a networked world. Therefore, marketing a good product has never been this easy.
CEOCFO: Are you able to continue the rollout the way you like? Is financing an issue for you or are you set for the immediate future? Mr. Ambati: We have not had much difficulty building a credible story and focusing on the fundamentals and focus on building a great team and happy customers. None of this is possible without the fantastic team we have built, which is comprised of some of the valley’s greatest engineers that I have wanted to work with for the last two decades. Once customers have tasted the quality of what we do and the speed at which we do it and the scale at which it brings, we are ready for the future. Building something good is always hard. Building a great thing is always hard. The rapidity with which we are able to build both our math teams and our engineering teams and our business teams has always been a positive surprise for people in the industrial community. They are seeing that we are able to maximize every dollar that has been invested in the company. We are reasonably well funded for the next year. We look forward to making the most out of the opportunity that is at hand, so we can build a fundamentally solid cash flow business for the next decade.
CEOCFO: Why pay attention to 0xData today?
Mr. Ambati:
0xData is
changing the way we have looked at decision making and predictions.
Historically, math and statistics were used to justify a decision that has
already been made in someone’s mind. What we are bringing is the latest
cutting edge predictive analytical algorithms to work your data & bring
surprising insights into your business. Most decisions are emotional. Humans
have feelings – which is great. Humans are the final frontier in big data
and decision making. Machine learning can deal with detail and machine
generated data. Our product H2O levels the playing field for data science:
algorithms that were predominantly the realm of sophisticated companies such
as Goldman Sachs, Amazon or Google is now open source and accessible. We
have brought advanced algorithms to the masses. This will unlock innovation
into so many fields. We think that data science is not rocket science.
Ordinary people will be to extend and apply this through Applications. Data
is the new Oil. Mining the data for outliers and insights is what 0xdata is
about. |
||
|
|
||
|
|
||
|
Big Data Analytics, 0xdata, Technology Companies, CEO Interviews 2014, Big Data Predictive Analytics and Machine Learning In-Memory Platform, open source H2O, Recent CEO Interviews, Technology Stock, the world's fastest in-memory platform for machine learning and predictive analytics on big data, running advanced algorithms such as GBM, GLM, PCA and RF users can get to interim and final results quickly to help them make better data-driven decisions faster, 0xdata Press Releases, News, Companies looking for venture capital, Angel Investors, private companies looking for investors, technology companies seeking investors, big data analytics companies needing investment capital |
||
