I train and evaluate multimodal AI models on image and video. I review and score model outputs across games, 3D assets and real-world footage against detailed quality guidelines, and write the feedback that makes them more accurate. I also create high-quality reference images and videos with Grok to raise the bar for generated content.
Eight years of editing, motion and photography is what I bring to it. I notice the frame that drifts, the light that doesn’t match and the motion that feels wrong. Specific projects are confidential, so this page describes the work in general terms.
To catch subtle defects between pairs of AI-generated videos in high-resolution footage, I built my own browser-based comparison tool.