Leading the Application Lifecycle team at Databricks to develop foundational building blocks and high-level abstractions for feature development for various high-scale AI and ML usecases including Model Serving, Feature Serving, and Vector Search. We tackle complex challenges - from building reliable RPC client-server frameworks, custom proto code generation for streamlining the adoption of Databricks API design standards and providing high-level abstractions to integrate with multi-cloud infrastructure.

Currently working on various bets in the Environment management team ranging from auto-restarts for services, host management, RDS management, migrating services to kubernetes etc.

Environment management team, Production Infrastructure : Design and maintain distributed daemons and micro-services used to manage hundreds of services, applications, and database systems for dedicated Gotham and Foundry installations running on VMs(in AWS/Azure), on-prem or in Kubernetes. Created services that enable continuous upgrades, service discovery, PKI, config, lifecycle & secret management as well as health status reporting for monitoring/alerting. A small peek on my work can be found here: https://medium.com/palantir/a-sky-full-of-clouds-218b9db3f735.

Orchestration team, Foundry platform: Designed and built scalable micro-services for scheduling and running efficient builds of highly-connected data pipelines involving 100s of datasets in Palantir Foundry. Mentored new hires and conducted technical interviews for engineering candidates. Replaced and migrated from a legacy system for quality and performance reasons.

Worked on the data systems team responsible for developing and maintaining the transaction API used by all applications to persist stateful information thereby abstracting away databases like Cassadra, Oracle or Postgres.

Architected and implemented a throttling system to ensure fairness in a multi-tenant environment.

Teaching assistant for Mathematics III, Computer Systems Organisation and Computer programming in different semesters.

Ads Delivery Insights team at Facebook HQ:

Built ML classification models for advertiser alerts and opportunities including real-time data retrieval pipelines.

Launched the model in advertising facing product by designing and implementing additional UI component for advertisers to get an overview of their Return on Ad spend and alerts on how to improve spending using the model.

Aditionally, implemented a history view in the Ads Delivery Debugger that helps browse and graphically visualize the various events/actions for any Ad entity and measure impact of the model.