My Engineering Journey with Team Bengaluru @ Guidewire
By: Satheesh Sudhakaran, Senior Software Engineer
Developing large-scale distributed system projects and learning new tech stacks with both the agility of a startup AND the financial backing of a great public company.
I recently celebrated my first work anniversary at Guidewire Software and wanted to share my experience with others. Guidewire may not be as well known in India, but is actually the world’s leading cloud software provider for the property & casualty (P&C) insurance industry.
Before Guidewire, I had 3 long stints developing software for larger companies, so this time I was looking to work in a smaller organization. With Guidewire, I found the perfect opportunity, having the agility of a start-up, but with the financial backing of a public company that has a $5B market cap.
I was lucky enough to be the first hire into the Cloud Data Platform (CDP) team within the Product Development Center being established in Bengaluru, India (located in the Prestige Exora Business Park on Outer Ring Road). It was challenging at first and had a steep learning curve because Guidewire’s data platform runs natively on Amazon Web Services (AWS), which along with Big Data, Kafka, and Kubernetes (K8s) were all new to me. But in the end, I was up for the challenge and sailed through it with the help of my colleagues and manager based in the US.
After some initial onboarding and technical orientation, we started our first project building a new CDP Orchestrator service that integrates the data platform with Guidewire’s core business applications (called InsuranceSuite). Built using a microservices architecture on Spring Boot and deployed on Kubernetes, this service orchestrates the provisioning of Data Platform resources when an InsuranceSuite application is deployed in the cloud. It was an interesting project because it gave me deep knowledge of how these two platforms (core & data) really work. It was also crucial to developing my understanding of the need to enable full automation and integration between the control planes of these two platforms.
Once started, we were able to complete a minimum viable product (MVP) implementation in 3 months and successfully deployed it to production. This involved collaborating with 6 different pods (which are small, cross-functional teams) within Guidewire Product Development & Operations. Our pod/team name is “Mysuru,” and it was a rich experience working with people not just in Bangalore, but in Kraków, Poland as well as the US cities of Exton and San Mateo.
After delivering the MVP, the product management team came up with an expanded set of requirements to make the service even better. This required not just orchestrating resources on the backend but providing some much-needed administrative visibility in the Guidewire Cloud Console (GCC) interface. This allowed the team to get an understanding of how GCC was put together using Guidewire’s Jutro Digital Platform (where each GCC add-in is enabled as an embeddable and reusable “micro frontend,” which was very cool!).
Other requirements included increasing the scale of the CDP Orchestrator data store that was originally designed into the MVP. This led us to adopt a single table approach based on Amazon DynamoDB, which is a serverless NoSQL database designed to run high-performance applications and services at scale. This not only provided the performance and flexibility we needed but significantly reduced the cost compared to a traditional relational database approach.
The following diagram provides an architectural overview of the CDP Orchestrator that was ultimately implemented by the team:
Another interesting problem we solved as part of the project was optimizing the delivery of streams of raw data to Kafka. To do this we used a Debezium source connector for Kafka Connect and pushed the data into Kafka in CloudEvents format. We also used the capabilities of Kafka Connect and Debezium to build a different plugin to intercept Kafka Producer messages before there were written to Kafka. Enabling these raw data streams and interceptor capabilities helped to establish a solid foundation for data integrity and auditing.
The high-level architecture of this work is shown in the following diagram:
Beyond the work itself, the high level of trust Senior Management provided to new teams like ours was both a pleasant surprise and very gratifying overall. All the projects we worked on were developed from scratch, and we were given complete responsibility — with guidance of course — to execute as needed. We followed agile methodology with ongoing sprints ending in production deployments every 2 weeks.
After my first year on the job, I am delighted to be part of Guidewire which really delivered on its promise to help me learn new tech stacks and how to apply them in real-world use cases. Within a year I got my hands dirty with Debezium, Kafka Connect, Kafka, and a bunch of AWS services like Lambda, ECS, S3, Cloud Formation, and many more. Career-wise, I began my journey as an individual contributor and currently taking the additional responsibility of leading a pod/team too.
As I said at the start, Guidewire functions like a startup but with the insight and maturity that comes from having worked successfully in the P&C industry for over 20 years. I believe this is a rare combination that is worth exploring for others as well.
Check out our previous blog to learn more about Satheesh’s team — Introducing Guidewire Data Platform: an enterprise-grade, internet-scale, and cloud-native Big Data platform.
If you are interested in working on our Engineering teams building cutting-edge, cloud-distributed systems and large-scale data platforms that enable a wide range of AI/ML SaaS applications, apply at https://careers.guidewire.com.
