Praneeth Katuri

- Architected and implemented a full-stack SaaS analytics platform from ground up, designing the end-to-end system architecture including authentication, authorization, and user management layers. Engineered a petabyte-scale real-time data platform using Databricks Delta Live Tables and Apache Pinot, enabling seamless ETL operations, materialized views, and interactive dashboarding capabilities with sub-second query performance. - Led the development of an autonomous dashboarding AI agent that interprets natural language analytics queries to dynamically generate and optimize visualizations. Engineering the agent’s reasoning layer to perform context-aware data exploration and metric computation.

- Redesigned the application release process to enable automated and parallel deployments, significantly minimizing dependencies. The new architecture cut release times by 70% and accelerated the dev-test cycle by 40%, significantly boosting team agility and productivity. - Spearheaded the strategic transition from Flux to ArgoCD across 15 production clusters, enhancing resource visibility and ensuring seamless integration with our core infrastructure. By capitalizing on ArgoCD’s self-healing features and its intuitive UI, we achieved a 30% decrease in incident resolution times, markedly boosting operational efficiency.

- Led the overhaul of Cisco’s secrets management platform using a Kubernetes operator to securely sync 8,000+ secrets with CyberArk Conjur. This transitioned password rotations from manual updates across 40+ clusters to a centralized system, greatly enhancing operational efficiency. Updates on the centralised secret store are now propagated to production environments in under 60 seconds, streamlining secret synchronization. - Developed a scalable microservice to efficiently process and export millions of critical data points related to client resource consumption. By leveraging Go’s concurrency primitives to implement a fan-out pattern, I enhanced data processing speeds by 60%, enabling seamless metric processing across multiple clusters.