Nafiseh Mollaei

I'm building Python-based agentic AI and machine-learning workflows for biomedical data at Northwestern University, the University of Chicago, and Loyola University Chicago. My work combines LLM reasoning, deterministic tools, imaging, scientific knowledge, and metadata automation.

I developed a LangChain-based agentic pipeline that automated contextual metadata generation across approximately 25,000 records using tool selection, structured outputs, validation, retry logic, and workflow coordination. I also built evaluation workflows with DeepEval and custom checks for relevance, factual consistency, reproducibility, and failure analysis, and created classification models that achieved approximately 85% automation across 50+ content categories.

Previously at Volkswagen, I led AI/ML and software projects spanning time-series analysis, computer vision, predictive modeling, and automated decision support. I build reproducible Python, SQL, Airflow, cloud, and data-pipeline systems that turn ambiguous research and operational questions into scalable analytical software.