Jana Kapr discusses how the Billigence Group works with H2O.ai.

 

Speakers:

Jana Kapr, CEO of Billigence Group

Read Full Transcript

Jana

Jana Kapr I'm a Billigence group CEO. What we do, we are business analytics company. What we deliver is data insights, data governance, cloud, data engineering and of course, data science. We've got offices across the world, as you can tell from my accent, I'm originally from Prague. We got office in Prague, Frankfurt, London, Warsaw, Singapore here in Sydney where we are headquartered from. And just recently we also set up San Diego in US.

 

Speaker 1

Terrific. The best companies have offices in Prague, as we like to say, because we have an office there as well. Tell me a little bit about the business problems or kind of the key challenges that you were hoping I could solve?

 

Jana

Interestingly, I studied engineering and water management at actually 25 years ago, ran models to monitor and do the flat and simulate floods to build dams to save people life, as she was talking about. So I work as my as part of my master's thesis with this Danish hydraulic institute model to build those and and really predict. And but it's very difficult I think.

Andrew McMullan on the stage when we had it actually you should probably got this out there, but what our solving is at 25 years ago was so difficult to run these models. It was very complicated to feed the information like hydrological cycle. You know, it took our geological profiles. We had to have lots of semi knowledge in order to make the model run well.

We only had a choice of one model because it was predefined. What took me two years to deliver? Now it would be a couple hours with H2O. And of course this is just huge advancements in this field. So I prior yes, you could build model. You could run the model. It took ages. That's why while you actually run it, it maybe took 12 months to get the results.

And this is not anymore. You know, we have it's so much timely nowadays and we have we could run so many different models. We could do comparisons. As I say, the board has changed.

 

Speaker 1

So having worked with H2O for a couple of years now, how would you describe the company to someone who had not heard of it before?

 

Jana

I would describe it as making really AI and data science easy and accessible to anyone who wants to learn it.

 

Speaker 1

And maybe that was great. Maybe just one more time and say to me, H2O, AI is just say the whole thing. But with that beginning of to me it's true it.

 

Jana

To me H2O AI is the seamless introduction to anyone what the power of AI and ML could be in any type of subject matter, expert, subject matter area.

 

Speaker 1

And how many data scientists or data engineers do you work with? Sort of give us a sense of the team or how many, how large it is.

 

Jana

Okay. So at Billigence, we are about 300 of us, but out of those 300 consultants, we have about 70% data scientists like data engineers or feature engineers. That would be of, I guess, our group. And we are constantly growing this, this area of ML ops and data science has been really scaling like this.

 

Speaker 1

Do you think you need to be a data scientist or a data engineer to use a tool like H2O?

 

Jana

You don't have to be a necessary data scientist, but you need to understand basics of data, I mean, statistical modeling, and you also need to understand what outcome you want to get out of your model. It's not that you can go and experiment, but at the end of the day you are looking for answers to what you want to achieve.

And I think basic understanding of statistics is really incredible. Like, I mean, it's necessary, not necessary for anything else.

 

Speaker 1

You would like to mention, anything I forgot to ask about that you think is important.

 

Jana

I think what's probably really important is that for anyone should not be scared of ML and there's a lot of buzzwords. Yeah, it's it's really we have, we used to report on the past information what we are now doing trends and we are doing bit more future data science, which means that we can actually really see what path we should take and what should be more efficient for us, or which path should be more efficient and more productive for us to take in whatever way or whether we if it's we want to increase revenue, we want to make sure we catch people who basically want to commit fraud or fraud, you know, whether we should offer a better products which are more suitable to the certain customers or whether we should or cost optimize certain certain product as well in order to get a higher sales numbers. There's so many different facets of of the future prediction of things and I think it's a big area of of what a lot of people didn't even think of, because everyone lots of people do gut feel even in a business decision making.

I think what I love about ML and AI, it gives us the substance to support some of that gut feel. Yeah.

 

Speaker 1

I wonder also as you talk about making more money, being more efficient, can AI and ML also be used for causes like preventing floods or climate change? Or does it have to be a business use case? Or can it be use?

 

Jana

That's that's the beauty of AI and ML I told you at the example I was doing, you know, with flat prediction, etc., what the new generation to allows us to is actually make the world around us safer from the disasters we could predict, you know, not only floods, but the climate changes. And that's I think it's beauty not just commercial side but that you really that doing good stuff for everything around us that our world is better so definitely climate change definitely water definitely.

Jana

And like one example in Australia we are so behind in flood management and predict prediction.

 

Speaker 1

Yeah, we talk a lot at H2O about democratizing AI, a lot of companies talk about democratizing and I wonder what that term means to you.

 

Jana

And your own. I think anything around democratization is, is to have it accessible to a lot of people. So I when I look at the organization, when we say democratizing data, it means we democratize access to the data. So we democratize access to the AI. That's why we teach the end users how to use the AI and we give them the tools to be able to use it.

Jana

And and that's for me is democratization.

 

Speaker 1

And why is that important?

 

Jana

I think the end users are a number of times those SMBs who need to use it in their day to day life. I guess it's all about giving decision making information to those end users. Yeah, it's not about us data engineers or data scientist. We are not there to at that point, whether it's a customer or whether it's a major decision where to build this dam or do this, you know, to prevent this flood.

Those are those end users and them, you know, that's why we need to make the AI much more accessible to these guys that they can tweak the models, which were well prepared. And I and I just laugh that when even sea level executives can understand, you know, and go and look at the model and try playing with it.

And what I love about H2O, that user friendliness of it, that you could actually explain it to people, you could you could prepare things for them and afterwards they could tweak it to the next level and use it for their day to day decision making.

 

Speaker 1

Any future plans for your data science efforts and how how does H2O play into that?

 

Jana

Okay, we have a lot of plans and there's lots of still I think we just scrape the surface. Yeah, I think the key is now is really how we integrate all the cloud technologies, like whether Snowflake, they create a etc. with H2O in order to match that democratization of the access and scalability of the data that we can create.

Lots of reusable feature set. So where we see a big move is how I could have reusable features which can be actually used by those and users using H2O that they don't have to develop it because that slows down the whole process. And of course not everyone can develop a feature. Yeah. So I think this is a big feature.

And the other really area, it still takes too long to get the models to the production, even like we see CBA as the front runner, but they still lots to do to actually really make it fast. And and I think that whole process of ML ops they still are so much which we can do better in order to make it faster, to make it more accessible, to get organized.

It's I see so much opportunities. We just I see it just like next five years. This would be the focus for us as a company, as diligence. And we want to really work very closely with H2O on this journey and of course getting all the smart data scientists involved and then users and good sponsorship from our clients that we can go of that journey with those early adopters.

 

Speaker 1

Anything else you'd like to add now?

 

Jana

I just love H2O, and I love H2O and Snowflake Combo, and I can't wait to introduce it to my clients.

 

Speaker 1

Thank you so much. That was great.