Muhammad Usman

Detail-oriented AI Data Annotator and LLM Trainer with 6 years of experience in data intelligence, catalog operations, and digital workflow optimization. Skilled in evaluating model-generated outputs, taxonomy mapping, RLHF pipelines, and error trace documentation to improve model accuracy and factual consistency. Proven expertise in translating complex operational criteria into structured data and high-quality guidelines for AI model training.

AI Data Annotator & Model Evaluator - Self-Initiated

(2024-01)

• Annotated and validated over 5,000+ complex data points and e-commerce product listings for LLM training, achieving a 95%+ accuracy rate through precise taxonomy mapping.
• Evaluated model-generated descriptions, reasoning steps, and search responses for logical consistency, bias, and hallucination identification.
• Documented error traces and created detailed failure analysis reports to help refine LLM prompts and guidelines.
• Designed QA checklists and annotation guidelines to streamline quality control across high-volume dataset pipelines.

E-commerce Operations & Data Specialist - Freelance

(2019-01)

• Managed catalog auditing, data formatting, and end-to-end operational workflows for 15+ digital projects.
• Created operational reporting dashboards using Power BI and Advanced Excel to track performance metrics and dataset accuracy.
• Streamlined inventory and workflow processes, improving output efficiency by 30% through structured tracking models.

Bachelor of Science - Data Science - Riphah International University (2020-01)