Feb 2024 – Mar 2025 · 1 yr 1 mo · Bengaluru, Karnataka, India · On-site

• I am part of the Data Platform at Uber which is responsible for managing the data lake and cloud lake for big data
• Worked closely with various distributed and large scale systems that form pillars of the data lake & cloud lake at Uber.
• Worked in the control plane to replicate Hive data from one region to another. It handles data at scale running millions of replication jobs per day and replicating peta-bytes of data per day.
• Build the features for replication of Hive data from one cloud region (GCP) to another cloud region (GCP)
• Worked on disaster recovery to handle partial and full failover of primary region to secondary region.
• Worked on migration of Hive data from on prem to cloud for cloud migration.

May 2018 – Feb 2023 · 4 yrs 9 mos · Bengaluru, Karnataka, India

• I am part of the Flipkart Cloud Platform team that builds and runs centrally managed services that serve the cloud.
• I have worked on multiple complex and distributed systems, and solved problems for scale at Flipkart level
• I have built Altair which is a centrally managed service(PaaS-DBaaS) that manages the entire life cycle of a MySQL cluster making it easier to plan, set up, operate, scale, and do more with a SQL database in the cloud. The key challenges in it were to make it policy-based and self-serve with APIs, automated workflows, and dashboards. Altair pioneered the managed service approach in Flipkart cloud platform and runs 500 MySQL clusters with 2000 MySQL instances in production in Flipkart.
• I am currently building Rigel which is a centrally managed service(PaaS-DBaaS) that manages the entire life cycle of a distributed NewSQL database (TiDB) deployment running on Kubernetes that is controlled by a k8s operator to provide multi-site, multi-cloud, and hybrid database topologies. The key challenges in it were to manage the NewSQL deployments on K8s-as-a-service, and to drive the adoption of standardized database cluster topologies for different high availability, scalability and HTAP use cases in Flipkart. Rigel is the first project in Flipkart cloud platform to use stateful K8s workloads for databases and hosts databases as big as 30 GB in size and 50 storage nodes.
• I am a Kubernetes developer and work on K8s operators and develop cloud services that run on Kubernetes in Flipkart and in GCP. I have contributed to the open source tidb-operator by adding features such as statefulset failure recovery (https://docs.pingcap.com/tidb-in-kubernetes/stable/release-1.4.1).

Aug 2011 – May 2018 · 6 yrs 9 mos · Bengaluru Area, India

• SPBM: Storage policy based management service in VMware vSphere product. My major contributions are - leading the effort to improve the product architecture for scalability, leading the effort to develop a unified framework for improving resilience and service availability along with other policy based management features in the control plane over VMware ESXi virtualization.
• ADP/vADM: A data center blueprinting and application dependency mapping product used for virtualization assessment services. My major contribution was pioneering in the delivery of robust security features and compliance.
• Chargeback: A virtual infrastructure costing application based on cost models and billing policies. I lead the development of an integration project to derive the billing cost of applications running in a cloud.

Jul 2007 – Jul 2011 · 4 yrs

• My major responsibilities include delivering high performance server side components, platforms, and end-to-end delivery of products that includes feature planning, complete software architecture, design analysis, evaluating third-party platforms and libraries.
• Projects that I have worked on include:
• Intelliview - a data visualization multi-tenant cloud platform which can provide detailed insight into past and live network access, analytics and business intelligence,
• eTIPS: enterprise platform that makes policy based decisions using identity-aware checks and machine health state and grants network access profiles to endpoints connecting to a network,
• OnGuard - agent software that resides on a computer and controls and manages state of network interfaces by collecting machine health information and evaluating it with a policy based decision-point server,
• GuestConnect: A hosted solution for provisioning visiting guests with joining policies and access restrictions.

Jul 2005 – Jun 2007 · 1 yr 11 mos

• Networx: A quoting and service ordering system built on top of a service delivery platform. I automated a user action flow to take a request in XML and pass it through multiple downstream services to create a sales order.
• RCR Project (Real-time Customer Record): A database repository to maintain real-time customer information’s image based on his day to day activity. I worked on a data quality check project used to compare and validate the real time data with a legacy service, and a bus service to construct and send recent image of a customer in the billing system.

Jul 2004 – May 2005 · 10 mos

• Message Store Platform: A storage platform for users’ postings and replies which was used as a backend for message boards and reviews and ratings. I wrote performance benchmarking scripts to outlay DB throughput in a web deployment and intelligently identify first bottlenecks in the multi-tiered architecture.
• Job-Mingle project prototype: A Referral Networks project built through integration of a social networking product and a jobs portal. I worked on a mash-up portal created by pulling out customer information from two different portals and presenting them in a combined structured way.