At Adobe Systems, I built a fault-tolerant async ingestion worker for ML ingestion using Java, Spring Boot, Kafka, and Redis. It supports crash-safe, idempotent processing across pod failures and redeployments.
I designed distributed job state and retry mechanisms, and scaled processing to 100 concurrent job threads across 20 pods and partitions. I also built a Kafka-based deletion pipeline that reduced worst-case cleanup latency by 5×.
For Acrobat Study Space and Generative Presentation GA, I designed and shipped REST APIs for AI-generated content lifecycle management. I also built a citation storage layer with six CRUD APIs and optimized listing latency from p99 2s to 900ms.
Earlier, as a Software Engineer Intern at Adobe Systems, I contributed a Watch feature for shared documents and APIs to add, remove, fetch, and notify watchers. At Indian Institute of Technology Delhi, I worked on xv6 scheduling and a DistilBERT project for hate-speech localization.

