Kunju Perath

• Shipped critical features addressing reliability challenges for large-scale cross-region customers using 5K+ total nodes. • Built distributed observability pipelines with OpenTelemetry, providing unified event, audit, and availability telemetry, cutting average detection time for control plane issues from hours to less than 5 minutes. • Spearheaded onboarding of DOKS to ATL1 (new Atlanta cloud region), navigating cross-dependencies and critical API incompatibilities to support the first GPU-focused DigitalOcean region in less than 1 month. • Implemented a multi-region Kubernetes operator that scales the DOKS control plane resources based on customer workloads, improving platform reliability and operational efficiency at scale. • Released DO Kubernetes v1.29 an effort that required ensuring DigitalOcean stayed compatibility an update to date with all core upstream kubernetes dependencies • Lead the release of DO Kubernetes v1.30

Responsibilities: • Backend Engineer: Developed critical microservices for DigitalOcean's Managed Kubernetes solution, used by businesses worldwide. • On-call: handled 30+ hours of high priority support issues from VIP customers. Impact: • Contributed to the successful migration of all legacy customer clusters to the new architecture by writing critical gRPCs and validation code that prevent inconsistent/outdated clusters from migrating. • Decoupled testing and deployment for a service from the main pipeline, reducing E2E test runtime by 80% and average build failure rate down by 50%, and as a result, saving 120 hours monthly. • Halved the ramp up time for new hires when setting up local dev by creating a detailed setup guide which consolidated all required dependencies and hosted weekly oce hours to support them with further issues. • Open source contributor to the DO Kubernetes Operator project used to manage DO resources from K8s.

Responsibilities: • Fullstack Developer: Developed Spring Boot REST APIs with React frontends for a large-scale automotive data platform. Impact: • Developed a React dashboard with Chart.js to centralize and display KPIs from multiple sources, saving 120-180 person-hours per month and increasing data accessibility for non-technical business users. • Extracted a key feature from a legacy app and developed a Go microservice with a UI that’s easily embeddable in multiple apps, resulting in a 75% reduction in development time and saving over 500 person-hours. • Wrote a Kubernetes cronjob that syncs Elasticsearch databases in dierent environments, reducing ETL usage by 50% and keeping processing time at 90 hours monthly even as new environments are added.

Did research on scalable dashboards as microservices and built a prototype dashboard using VueJS and NodeJS.

Undergraduate Thesis: Using Apache Spark to handle distributed text classification using a modified kNN algorithm Relevant Courses: • Analysis of Algorithms and Data Structures • Object Oriented Design • Operating Systems • Networks • Foundations of Computer Science • Intro To Machine Learning • Distributed Systems