Lexicon — live lecture captions for deaf and hard-of-hearing students, built on AssemblyAI streaming. Generic captions mangle the exact words that carry a lecture, so Lexicon takes the course glossary as an input and adds an ask-panel that answers questions from the last ten minutes of transcript, with a timestamp citation you can jump to. Working solo and shipping regardless — open to a teammate with accessibility or frontend experience, or anyone who's worked with deaf/HoH users and can tell me where the UX is wrong.