Radhika Fichadia

About

Backend/DevOps-focused engineer with healthcare and financial systems experience, including an ICU readmission risk platform delivering real-time ML scores via a secure FastAPI service (PyTorch model serving, PostgreSQL, Celery/Redis) deployed on AWS with strong observability. Has hands-on Kubernetes GitOps delivery (Helm, ArgoCD, HPA) and has supported a JPMC on-prem-to-AWS microservices migration using phased validation and blue-green cutovers, plus Kafka/Avro streaming for real-time transaction processing.

Experience

Education

Key Strengths

• End-to-end ownership of secure, real-time ML inference APIs (FastAPI + PyTorch)
• Productionizing ML for low-latency inference (standardized preprocessing, in-memory artifact caching, CPU-optimized deployment)
• Strong API/data contract rigor via strict schema validation (Pydantic) to prevent mismatches
• Cloud-native containerization and deployment on AWS (ECS/EKS) with monitoring via CloudWatch
• Kubernetes delivery maturity: Helm-based deployments, HPA scaling, GitOps with ArgoCD, and rollback governance
• Migration execution from on-prem to AWS using phased approach and blue-green cutovers
• Real-time streaming pipeline design with Kafka/Avro; handled peak spikes via autoscaling and idempotent consumers
• Observability-first approach: custom metrics, alerts, dashboards, and tracing to surface issues quickly