Building Data Infra

Using Graph Neural Network to predict a referral link between a patient and specialist. Currently working on model explainability using captum.

• Managed multiple microservices in a production environment, ensuring their seamless orchestration, and reliability.

• Collaborated closely with the platform architects and product team to drive low-level design and feature prioritization, leveraging cross-functional expertise across multiple projects to ensure alignment between system design and product expectations.

• Improved the performance of a production inbound microservice by a measure of 26X from 0.6 TPS to 16 TPS on a per node basis, by solving for data serialization latency b/w partitioner and reader layer of the batch component and implementing a solution to distribute events more evenly across nodes.

• Collaborated in developing multiple event-driven spring boot-based microservices assisted with frameworks like Apache Kafka for event publishing and consumption b/w services, and Redis for server-assisted client-side caching.

• Contributed towards setting up contract testing framework and standards to be used between microservices, intended to discover bugs early, and to reduce dependence on integration testing.

• Developed an S3 storage-based utility library, which is used by all microservices built to perform upload and download of multipart files with/without encryption to/from ECS Storage.

• Built Spring Batch applications for data migration and purge jobs assisted with file watcher services for inbound file arrival discovery using Java filesystem APIs.

• Created bash shell scripts to handle file management, file purging strategies, and application deployment on Linux servers.