Apr 2023 – Present · 3 yrs 2 mos · Bengaluru, Karnataka, India · Hybrid

• ● Working on development of an inhouse petabyte-scale Global Observability Platform which can collect, process, store and visualise telemetry(traces, logs & metrics) data. Building it generic enough to extend to application feature-logging or other system/business metrics related wide-events for realtime analytics and producing results for further OLTP decisions realtime. Built our telemetry pipeline using open-telemetry protocol, Kafka, Clickhouse and Grafana with an easy adoption curve, cost effective and highly available.
• ● Working in the backend engg platform-team for development of generic & centralised libraries/frameworks, wrappers/abstractions, services, platforms for streamlined infrastructure access, improved engg processes, best practices, enhanced developer experience, and efficient microservices development across teams.
• ● Worked towards evolving the ShareChat’s queueing infrastructure, data compression, migration from GCP’s PubSub to Kafka/Redpanda. Did a thorough Redpanda POC wrt our huge scale, high latency sensitive data, our traffic patterns, performance, fault-tolerance, availability and accuracy. Found bugs and got them fixed, created roadmaps/plans for all teams for effectively migrating to new queuing system in the org, developed generic & centralised libraries and tools for devs to have enough incentivization, best practices and easy to migrate. Saved 70% of the cost of our queueing infrastructure.

Aug 2021 – Mar 2023 · 1 yr 7 mos · Bengaluru, Karnataka, India · Hybrid

• ● Worked on developing a generic LiveMigration framework to migrate tables data (>100TB) from our primary DBs (GCP’s Bigtable & Spanner) to our new vendor-agnostic open nosql DB (ScyllaDB) without downtime and >99.99% accuracy. The framework involved platform libraries to orchestrate the live migration of traffic from various services on Kubernetes without disturbing business logic and central control, also involved the use of Kafka to mitigate the errored rows while live migration and also use of Google dataflow batch processing jobs to bulk export the existing data to the new database.
• ● Developed a compute autoscaler for Google Spanner database to ensure consistent read/write latency during our peak traffic times and festive traffic peaks, also enable better efficient utilization of compute and respective cost savings during the off-peak times. The autoscaler highly configurable wrt various database metrics and is deployed in production for all our Spanner database instances.
• ● Worked towards developing a disaster recovery mechanism for our primary databases for ensuring a lower RTO and RPO in-case there is any outage in our databases by a human/developer error or because of any infrastructure issue. The task involved developing a watcher component which can detect issues quickly and can redirect the traffic to healthy/backup database instances, the task also involved development of automatic regular backups of database for to ensure protection with minimal RPO with respect to any corruption in data.
• ● Worked in the backend engg platform team and responsible for development of various generic & centralised platforms, wrappers and libraries for simple, generic and reliable infrastructure access, promoting developer experience and facilitated reliable & performant development of microservices across teams.