Performance engineering team for the #1 Analytics product in the world (Salesforce + Tableau)

Architect and product lead

Developed Edge computing platform running on Kubernetes with software LB and deployed on PoPs. Features - Session acceleration, Dynamic caching, Security pipeline, Datapath anomaly detection. Interaction with Frontend, Tools and Operations teams to rollout the solution to the Edge. ~250ms network latency reduction to all the eBay user flows. Identify and work with teams/applications to prioritize onboarding on this platform. Work with their scrum teams to execute faster.

Data framework for aggregation, enhancement, analysis and visualization from heterogeneous channels.

Developed a data abstraction framework and data enrichment layer to ingress logs and metrics from different sources. Coordinate with multiple teams to Integrate this framework into their applications.

This provided eBay with unprecedented insight into the traffic pattern and user behavior and its relation to the business metrics. The analysis and visualization enabled the ops and execs decision making with near-real time insights and trending for security, traffic shaping and ISP irregularities. Exposed this as a service to be consumed by internal customers for further analysis.

Dynamic Caching on the Edge – phenomenal improvement in user experience

Identify the static and user specific (personalization) content in the application response and orchestrate a flow to cache and quickly deliver the static content to the users (TTFB improvement by 500ms) and augment the user specific content with a call back to the Datacenter in the context of the same request. The response is delivered seamlessly to the user in multiple chunks. This orchestration is implemented as a filter in Envoy proxy

Cloud optimization – Instant flex up/down, Optimized load balancing, Resource based allocation

Collaboration with Hardware Engineering and Datacenter teams to enable scheduling of applications based on application characteristics and optimized load balancing algorithms. This effort reduced the cloud infrastructure capex and increased the resource utilization by 33% increased the applications performance by reducing the neighborhood noise for services co-hosted on the same node in the cloud.

ESB and API gateway – Rearchitect the API gateway framework to handle all the Native app traffic.

Implemented an open source based horizontally scalable architecture catering to microservices and legacy applications. This is handling more than 60% of eBay traffic providing deep insights and visibility.

JVM Monitoring and Diagnostic tool. Enabled administrators to diagnose performance problems in Java application in the production environment without having to reproduce them in the test or development environment. It provided in-depth visibility of JVM and java application activity with low overhead. Integrated with Oracle Enterprise Manager and enabled diagnostic data collection from different vendors.

JVM Tuning and performance improvement of eBay’s kernel framework and applications.

Identified scalability bottlenecks, achieve the optimal QoS and help with capacity planning. This resulted in reduction of capex (infrastructure) and opex (increased availability and reliability) costs for critical applications