Deterministic canvas render using the nominee's real uploaded photo.
1080×1350 (4:5). Her real LinkedIn photo is composited into the copper laurel template. Name text is rendered deterministically — no AI, no misspellings.