Worked on the primary high-volume tweet backend that serves all the tweet read (tens of millions per
second) / writes (tens of thousands per second) traffic. Worked on making Tweets simple to extend
& serve at scale, to enable product teams to iterate quickly on a new feature. I actively contributed
to Edit Tweet, one of the most sought features on Twitter. Some of my tasks involved handling the
aggregated fields in the edit chain; operations associated with edit versions like retweets, etc., and
providing backward compatibility for the edit features (Project length: 3 quarters).
•
To reduce the run-the-business costs, migrated critical legacy link shortener infrastructure to the
internal serverless platform strato (relatable to AWS Lambda). It involved understanding the legacy
code and breaking the service into components that can be run together on this self-serve platform.
This effort required thorough comparison testing, gating with real-time monitoring, detailed runbooks,
etc.
•
Handled rotation-based on-call responsibilities which included tackling customer queries & Addressing
their high-urgency demands, monitoring in real-time the deploys of various tweets & URL-shortening
related services, and investigating the alerts if any of the services faced operational issues or the deploys required any sort of manual intervention.
Worked on Nvidia Cloud Platform, Nvidia's homegrown cloud management platform that manages the data centers across the globe and is built using custom Kubernetes components (written in Golang).
•
Centralized Virtctl VNC - Extended the VNC module of virtctl component of the Kubevirt project that remotely connects to the vmis using a serial console.
•
Evaluated the flow of data bits from virtctl binary to the virt-api server and the virt-handler.
•
Examined various security aspects and performance metrics for scalability, the commit is in pipeline to be merged with the upstream codebase.
•
Prometheus and Grafana - Added Prometheus metrics to existing custom Kubernetes components and created various Grafana dashboards for insightful visualizations.
