Nicholas Yardich

Architected advanced data pipelines for fintech enterprise analytics using Python microservices and metadata frameworks to enrich data definitions, enable intelligent keyword discovery, and reduce report build time by 35% across 12 business units. Built scalable backend services with ETL/ELT pipelines and star-schema models to transform structured and unstructured sources while lowering query latency by 40% for regulatory reporting, supporting 15 source systems. Helped shape engineering best practices through CI/CD adoption, automated testing, observability, and code reviews—delivering reusable semantic models and reducing deployment lead time by 50% across 12 delivery teams. Mentored engineers and led architectural decisions, aligning observability with model evaluation and improving review quality by 30% across 8-person squads.

Built LLM-based intelligent automation services with orchestration and agent frameworks, integrating distributed services and APIs to automate data enrichment, field mapping, and enterprise data transformation—cutting manual review by 60% for 20 datasets. Optimized Spark and Lakehouse workflows on AWS by designing reusable dimensional models and star-schema generation, improving analytics performance by 45% while reducing storage costs by 25% across petabyte-scale environments for risk analytics teams. Architected intelligent data discovery and recommendation using knowledge graphs, semantic technologies, and vector search—helping analysts locate trusted datasets 3x faster and increasing self-service adoption by 55% across 9 product teams. Partnered with architects and governance teams to establish performance, security, and observability standards via release automation and infrastructure-as-code, reducing incident volume by 40% for 14 monthly releases.

Led LLMOps and MLOps practices for semantic search and natural language query features, integrating model evaluation dashboards to reduce answer errors by 35% across 5,000 weekly active users for students and administrators. Developed data integration workflows across student information, learning activity, and assessment systems, applying responsible AI checks to improve report accuracy by 48% while protecting privacy across 12 districts, 3 campuses, and 8 schools. Created scalable APIs and distributed analytics services using containerization and event-driven patterns to help teams deliver self-service dashboards 50% faster, reducing support tickets by 25% across 7 product squads and improving release confidence by 30% over 3 monthly cycles.