SpringML in 2020

Welcome to another episode in SpringML’s Podcast Vision 2020.

Today’s episode features an interview with SpringML’s Chief Technology Officer and  Co-Founder Girish Reddy.  He shares his perspective on analytics and AI ML in 2020.

Episode Transcript


Megan
– Girish, great to have you today. I want to ask a few questions about 2020 analytics and AI & ML this year and what’s coming and your perspective on how SpringML is changing and growing with the projects that we were working on in 2019. What’s changed?

Girish –  I think that in terms of Machine Learning, the customers that we work with are transitioning from more experimental to actually deploying models in production so that they can see and leverage the results of those models. In 2020 it’s no longer in an experimental or research phase, but instead, take on use cases that can help them meet some of their business objectives. So that means for us it’s an end-to-end life cycle of machine learning. It’s not just piloting and prototyping a model and looking at the accuracy but rather where we deploy a particular model in production. So what that means is when we talk to customers, we want to understand what their motivation is in embarking on a specific project and helping them think about things in the right way – guiding them.

We’ve done things in the past so that we can help take a model to production and not realize midway that perhaps that’s not the right use case. We take the time upfront to discuss use cases with a customer via workshops. These are typically whiteboarding, brainstorming kind of sessions, but zeroing in on those use cases, which have the maximum impact on their business.

Megan – When it comes to GCP or Google Cloud platform in our partnership with them, what’s changed is our offerings for 2020 as compared to 2019?

Girish – We’ve had the good fortune of implementing these cutting edge solutions around machine learning & also in data analytics, data warehousing for large customers. We can bring the best practices, and things to watch out for when embarking on an ML project. We can guide customers upfront during the initial design discovery phase, identity things to look out for based on our experience. And I think that’s an invaluable piece of expertise that we bring to the table. 

Megan – We’ve also expanded into more significant workload migrations, things like SAP migrations, extensive data warehouse migrations, as well as consulting and supporting our customers in assessing which workloads to move to the cloud. Could you speak to those additional services and how we’re framing them in 2020?

Girish – There are a few critical killers around data warehouse migration and application migration for customers. If you look at that piece of workload, it is automatically other things such as VM migration, application migration, Kubernetes deployment, an application that might be on-premise into the cloud using Google Kubernetes. All of these are, and I see them as being tied together. They cannot be a silo in nature. So when we deploy a large Data Warehouse in the cloud, we know that there are surrounding up. Station surrounding clients surrounding APIs that will need to migrate to the cloud as well. So we look at that holistically and help migrate the customers identifying all the dependencies to the cloud, making sure that the key requirements are being met around performance, security, networking, etc.

Megan – You mentioned networking, and I’m curious. I know that we have a customer. That’s just coming onboard very soon here with a very large network infrastructure project. Could you walk us through the type of work that we’re doing there? 

Girish – We need to look at security networking and how the firewalls, what kind of firewall rules need to be set up, who has access to what kind of data etc. So these are some things that when said, even if you’re doing a data warehousing project being able to talk to what kinds of applications, what kinds of clients would need access to that data understanding those requirements and then setting up the networking and security infrastructure to meet those requirements become part of the implementation. And some projects might have more networking and security requirements where we are migrating applications to virtual machines and such. But, overall, we have a methodology in place that helps us to go to each of those are different areas of discovery namely security, networking, user access, scalability, performance, disaster recovery, all of those things coming to picture as part of any project implementation.

Megan – When it comes to Integrations and thinking about how we leave all these different applications that are producing data, transferring data. Do you have any tidbits for our customers and for the audience here & our partners on how to think about immigration in 2020?

 Girish – It has been a pain point for customers for a long time. Integration within the realm of data warehousing and machine learning specifically. We generally take the approach of accessing data from source systems and dumping it into a data warehouse like BigQuery before doing further optimization to meet the specific customer demands within Google Cloud dataflow being one of the primary ones. We also use cloud composer for orchestration and integration of various data sources into a target system like BigQuery using these tools becomes relatively from the technology point of view relatively straightforward. Now the areas to look out for are around again security. Are we bringing in a PII data that may require additional encryption and so on what kinds of access privileges do we have to access applications that may be residing in customers on-premise data warehouses. Usually, there’s a lot of planning in terms of accessing those systems. 

Understanding the data schema data dictionary etc. requires us to talk to the appropriate data owners. So, more around planning and designing of the integrations around, understanding the data more than technical challenges is how I would say integrations would work out. 

Megan – That’s fascinating and a great selling point for our services and for AI & ML. I would think to help address some major headaches in the industry for many years. Well, Girish, It is an absolute pleasure chatting with you, and I can’t wait for our next interview where we can pack some of these customer’s stories. Thank you.