RAJ KUMAR

About

Backend engineer at Discover who built and scaled Python/Flask services for a card dispute resolution platform, tackling long-running external network validations with Celery+Redis and delivering measurable gains (response time ~3s to <300ms; throughput +40%). Experienced in high-scale PostgreSQL/SQLAlchemy optimization (partitioning, read replicas, N+1 avoidance), event-driven systems with Kafka, and integrating ML fraud detection using AWS SageMaker/Lambda/ECS with clear separation of real-time vs batch processing.

Key Strengths

• Built Python/Flask backend services for card dispute resolution platform across multiple card networks
• Resolved long-running external validations/timeouts by introducing Celery + Redis async processing
• Strong SQLAlchemy performance tuning (indexes/constraints, lazy vs eager loading, N+1 avoidance)
• Scaled PostgreSQL with monthly partitioning and read replicas to separate analytical workload
• Integrated ML fraud detection using AWS SageMaker with Lambda orchestration and REST deployment
• Designed scalable containerized backend on AWS ECS with separation of real-time scoring vs batch jobs
• Implemented event-driven architecture with Kafka to decouple services and process events in parallel
• Implemented schema-per-tenant PostgreSQL multi-tenancy with tenant-specific configuration and connection pooling
• Quantified performance impact: reduced API response time from ~3s to <300ms and improved throughput by ~40% via background task optimization

Experience

Education

Skills

Languages

Certifications