At New World, I worked with developers on an enterprise data analysis platform with an LLM assistant. I built multi-source ETL pipelines for ERP data and developed a semantic layer that mapped each client’s columns to shared business concepts.
I also developed a two-step assistant that calculates figures deterministically with Pandas before using an LLM to explain them, alongside a quality-control agent that compares results against reference data.
As a freelance Data Scientist, I designed data models and ETL/ELT pipelines for retail and SaaS clients, and automated data preparation that reduced processing time from one to two days to under three hours. Earlier, as a Data Analyst at a Tech & Data consulting firm, I developed a document-research RAG assistant, unified multi-source reporting data, and maintained Power BI and Streamlit dashboards.
