I study how machines read imperfect traces.

Hello — I’m Ema. I work on machines that try to read difficult documents: noisy newspapers, OCR and HTR transcripts, handwritten archives, inscriptions, administrative records, financial texts, multilingual media, and visual cultural collections.

I care about what AI extracts from them — people, places, events, relations, narratives, uncertainty — but also about what it misses, invents, flattens, or quietly ruins while looking very confident. I come from both computer science and art, so I tend to treat documents as data, evidence, image, memory, and occasional trouble.

Outside research, I keep returning to photography, handmade things, gardening with variable success, films, series, mostly non-fantasy books, metal festivals, camping, and soup logic. I like clarity, good noise, dry humor, and people who get to the point without murdering the point.