AI Trainer – Multimodal Evaluation Handshake AI Fellowship (Remote)
Evaluated high-volume multimodal AI outputs by reviewing generated images and related visual components to assess model performance. Provided detailed technical critique focused on instruction following, person ID preservation, and the identification of AI artifacts. Performed expert-level assessments while complying with strict confidentiality protocols. • Assessed image generation quality against required constraints • Checked for preservation of identity/person ID in outputs • Identified and reported AI artifacts and quality issues • Documented rating-style feedback to guide model refinement