At Verizon, I engineer LLM-based multi-agent workflows, RAG pipelines, and real-time voice features that improve reliability, retrieval precision, and deployment speed. I reduced hallucinations through error analysis and prompt adjustments while building evaluation gates using trajectory evaluation, golden sets, and LLM-as-judge frameworks.
Previously at Northern Trust and UTA, I built production machine learning pipelines, optimized distributed data systems, and automated data workflows with Python. I bring hands-on experience securing sensitive claims data, integrating APIs, and measuring AI performance with evidence-driven evaluation loops.
