Linjia (Ella) Zhou

Prompt Engineer at Meta

Designed and executed structured test plans to evaluate LLM and multi-agent system performance, ensuring reliability and accuracy prior to deployment. Developed Python-based automated data-quality checks for analyzing annotation data, automating evaluation workflows, and generating key insights used by engineers to refine Llama model behavior. Reviewed code generated by GenAI models, identifying inconsistencies to improve overall robustness. Conducted quantitative analyses using Python and SQL-driven Sheets integrations, including rating metrics, distribution analysis, and system performance comparisons. Authored clear and comprehensive test reports summarizing defects, risks, and recommendations used by stakeholders to improve Llama model quality. Onboarded and trained new team members, serving as the primary point of contact for new hires.

Lead Data Quality Specialist at Cohere

Implemented detailed plans for optimizing various ML data, including code, research essays, government forms, and RAG responses. Managed team responsible for labeling, ranking, auditing, and correction of ML data, ensuring high quality. Recommended optimization opportunities to Large Language Models developers based on detailed analysis. Worked with cross-functional team members, fostering effective teamwork and continuous learning.