PAW 2019 Podcast Series

PAW 2019 Conference Series – Satish Pala – Indium Software

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 Satish Pala, Senior Vice President, Digital at Indium Software.

Find Satish Online:

Email: satish.pala@indiumsoft.com

LinkedIn

Indium Software


[00:02]

You’re listening to the Happy Market Research Podcast.  I’ve already screwed up this intro a couple of times. I apologize for that to my wonderful guests.  Right now, I’ve got Satish Pala, Indium Software. He is a market veteran, been in the industry for 20 years.  Welcome to the podcast.

[00:22]  

Thanks.  Thank you so much.

[00:23]  

We are live today at Predictive Analytics World and Marketing Insights World and all kinds of worlds.  I think there’s 12 shows. It’s a lot of shows. That’s right, yeah. From health care… Everybody’s using insights.  What do you see as one of the big trends in the industry?

[00:42]  

So, one of the things that I see is people having a lot of data, people thinking a data analytics solution will help them but have no direction or a strategy towards these concerns of theirs.  So they are trying to do what I call a prototype-based analytics. So, they would want to spend some money, figure out what insight could come up with that amount of data they have in a shorter cycle, maybe a month or two.  And then they figure out how it is going to impact their business. And based on the business impact, they’re trying to spend more, invest more in the scheme of data analytics where they can get impact, a positive impact on their business.   So, for example, if a company who’s having products related to customer, consumer, for example, electronics and they see the trend in electronics buying has reduced, they would want to analyze this challenge. They would want to analyze why the trend is so in terms of how the consumer behavior is.  So they would want to take the data that they have and analyze the patterns of consumer behavior and identify how they could fix it or, I would say, how they could work around these challenges that they’re facing. Having analysis on data insights on data can give them a picture of what they need to do in the near future.     

[02:04]

So, Indium Software, you guys have been around awhile?

[02:08]

Yeah, we’ve been around for almost 20 years.  And, as you said, we are veterans in the IT services.  We deal with quality solutions, services. Primary, I would say the portfolio that we solve, is at once analytics; then we have Big Data analytics, where in we help customers do data engineering, where we bring in all the data into a warehouse or data lake.  We give them business intelligence solutions; we give them visualization solutions; we also offer descriptive analytics solutions; we also offer advanced analytics, predictive-modeling-type solutions. We also help them integrate these solutions into their web portals.  So, one of the key challenges we have noticed is that, while they have done their analytics model on a dashboard or a visualization layer, they are trying to figure out how do I input this into my day-to-day life or day-to-day operations. So that is something we really focus on and help the customers with.       

[03:07]   

You know it used to be the case that data was far away from the person that needed it, and now it’s moving really close, even like integrating insights into the decisions that are the workflows of the practitioners that are inside of the brands.  

[03:25]

I don’t want to interfere in your questioning, but the thought you have is right.  People have identified analytical models; people have insights. But how do they operationalize these insights?  How do they put in the business…?

[03:37]  

I love that term “operationalize.”

[03:39]

So, we normally have an analytical model life cycle that starts with data preparation, modeling, model management, and operationalization of analytics.  So, this is something that we’re very good at. We take the insights and put it into the business process so that the business can see the value out of this model.  So that is one of the challenges where each business is trying to see how my model helps me. OK, I invested so much in analytics, but how does it help me in regular day-to-day operations.  Is there an automatic, seamless way of identifying the solutions for these challenges I have? Can these models be running a business like a regular day-to-day work instead of some data scientist coming to me with a report?  So that is the key. And the seamless integration of modeling into business processes is called operationalization, which we are very good at.

[04:30] 

That’s fantastic.  Do you have a favorite customer story?    

[04:35]

Oh, yeah.  So, I have a couple of stories, but let me focus on the one. So, let me focus on the one where we have helped them predict, I would say, anomalies on their production line or on production output.  We also helped them identify all the parts or the assembly line that they have, which can fail in a few weeks. For example, let me elaborate. This is a manufacturing company, and it manufactures a very niche product, which has advanced machines that are fabricating this product.  They have a challenge where they want to identify the anomalies in this product because the product is so niche, the quality needs to be optimum quality, right? So they wanted to analyze the data that is available with them to identify all the products or all the output of this assembly line that can be outside the specifications that they have.  So, we had helped them build an analytic model to detect the outliers, I would say. These are the anomalies. That is one, which helped them make the productivity higher, maybe make it, I would say, the assembly-production process was much more efficient than before.    

[05:56]

So they used that data in order to identify where the anomalies were taking place.  They could focus on improving it, whether it was like a lean approach or maybe a machine or…

[06:07]    

There were various things that they could do.  They could identify problematic machines that were fabricating these devices.  Or they could identify the process, by which they are fabricating. They could identify maybe a change to the product itself.  Let’s say you’re getting ten out of a million products. Is it the right number? Should I make it zero defect, zero outliers. So, that is something that is very important for them to have zero outliers because it’s a very costly product and niche product.    

[06:38]

Yeah, for sure.  

[06:39]

The second thing that we have helped them with is these assembly lines or these machines that fabricate this product, they have various sensors on top of them.  So, these sensors generate events. And these events, for example… I’ll just give an example, right? So, one of the sensors is a temperature sensor, the other one probably related to luminosity.  So, the outcome is that they want to identify how safe is my assembly line, how safe is my assembly station. And they have data available from these sensors. So, there are thresholds set for this sensor data that   if the temperature goes above so much Fahrenheit or the luminosity is above so much levels, which is safe zone, they would want to flag it, saying that this particular assembly station could fail in two weeks or three weeks.  Would you like to do preventative maintenance so that it is safe and you don’t have an accident later? So it’s pretty advanced because we had to bring in the IoT data. So this sensor data generates IoT data.

[07:44]  

Are you partnering with somebody to do the IoT?

[07:49]

No, we do it ourselves.  

[07:51]

It’s all yours, yep, right.  You have some kind of like sensor that you’ve installed on the…?  

[07:57]

OK, in this particular scenario, the sensors were already installed by them. 

[07:59] 

OK, got it.

[08:01]

We helped them ingest the data from these sensors into a data platform, which we built.  And then, we developed an analytical model to identify outliers, the anomalies. We also built a predictive model to identify when a part can fail or when a station can fail.

[08:15]

So it’s predictive maintenance is what’s happening.

[08:20]

Predictive maintenance is the key.  Two things: one is identify the outliers; second is predictive maintenance, which is very important because it is better you perform preventative maintenance rather than reactive where you have to spend a lot of money.    

[08:33] 

Yep, makes sense.

[08:35]

And this particular company built it for themselves.  Now that it’s working for them, they are productizing it, monetizing it, positioning it for their vendors.  So, that’s the nice story and we’re very proud of this because it really helped them improve their efficiency. 

[08:50]      

Indium Software, if somebody wants to get in contact with you, how would they do that?

[08:53]

So, you could go right into the web portal www.IndiumSoftware.com and then you will see key-members profiles available there.  Or you could go to LinkedIn or you could go to social media, Twitter.  We’re available everywhere, and we will contact you right away.

[09:08]

Satish, thanks for being on the Happy Market Research Podcast.

[09:11]  

No problem.  Thanks, it’s my pleasure.

[09:12]  

If you enjoyed this episode, please take the time to screen capture, share it on social media.  Special thanks to Predictive Analytics World. Really appreciate you guys hosting us. Have a wonderful rest of your day.