AI Content Evaluation & Fact Checking (System1)
I cross-checked AI-generated content against structured reference data to catch factual mismatches and improve editorial trust. I partnered with data scientists to tune evaluation rubrics for LLM outputs, ensuring annotation checks aligned with business rules and quality gates. I built Python validation scripts to spot outlier patterns, incomplete records, and inconsistent labels in daily ingestion feeds. • Compared source records, flagged assertions, and output summaries in internal QA tools. • Normalized research-style metadata to reduce duplicate entity matches in batch review runs. • Presented findings from weekly QA audits to stakeholders with actionable recommendations. • Used PostgreSQL, MySQL, and Python for traceable validation across datasets.