Your question is Embeddings for Medical Text Search. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
You're working on search over clinical text, where keyword matching often misses relevant notes that use different wording for the same concept. You want a representation that captures semantic similarity so related medical phrases can be matched even when the exact terms differ.
How do embeddings work, and how would you use them for medical text search?