Sowmya Kasu

About

Backend/ML engineer with healthcare experience at Kaiser Permanente building HIPAA-compliant Java/Spring Boot + GraphQL APIs integrated with Epic HealthConnect, including hands-on reliability/performance debugging using Prometheus/Grafana and resolver-level N+1 elimination. Also built an end-to-end malaria parasite detection ML feature (CNN/R-CNN) with evaluation, guardrails, and workflow integration, and has experience designing robust state-machine-based automation with retries, DLQs, and alerting.

Key Strengths

• Debugged and reduced GraphQL P95 latency by eliminating N+1 queries via batching/caching (DataLoader pattern)
• Performance tuning using Prometheus/Grafana metrics plus SQL profiling and targeted indexing
• Designed HIPAA-compliant healthcare APIs integrating with Epic HealthConnect
• Built end-to-end applied ML pipeline for malaria parasite detection (data prep through model evaluation and integration interface)
• Designed reliable multi-step workflows using state machines with typed error handling, retries/backoff, and DLQ-based replay
• Improved slow PostgreSQL dashboard queries via composite indexes, denormalizing current status fields, and materialized/cached aggregates
• Production readiness mindset for AI systems: observability, offline eval/regression checks, drift monitoring, fallbacks, and human-in-the-loop

Experience

Education

Skills

Languages