Roger Herzfeldt

Led design of Python microservices and LLM orchestration in Docker-based cloud deployments, improving document discovery latency by 38% while meeting enterprise scalability, metadata management, and security requirements. Architected star-schema generation with Spark pipelines and Azure/GCP integration, reducing customer query execution time by 46% and improving deployment performance via automated testing and metadata enrichment. Optimized ETL/ELT workflows with reusable integration components, accelerating marketing automation reporting cycles by 25% and aligning roadmaps with security validation. Mentored 5 engineers with technical leadership, establishing MLOps CI/CD and containerization best practices that improved backend scalability, LLM application performance, and security governance by 30%.

Designed event-driven credit risk microservices with Python and Azure using Docker containerization and idempotent consumers, increasing approval throughput by 41% while preserving distributed-systems reliability and audit logging. Implemented Spark query optimization and schema evolution in a lakehouse analytics layer, accelerating analysis by 35% with automated testing, security controls, and performance monitoring. Built LLM fraud detection services with Google Cloud, CI/CD, and Kubernetes, improving alert accuracy by 52% and supporting scalable fraud review workflows with observability and audit controls. Provided technical leadership and mentorship that increased release frequency by 2x and reduced incident recovery by 40%.

Architected AWS Lambda services and distributed query APIs for catalog search, achieving 99% uptime and 28% faster response via caching, read replicas, and semantic metadata governance. Optimized ETL/ELT pipelines with Spark, Docker, and GCP, reducing nightly batch latency by 37% while improving analytics throughput and enterprise security and schema governance. Established CI/CD deployment patterns using containerization, Helm, and OpenTelemetry, reducing release failures by 46% and improving performance validation across APIs. Integrated LLM recommendations and AI scoring with MLOps to deliver a 22% conversion lift from personalized commerce automation, with robust analytics and governance practices.