🧠 **How it works:** → The user writes their perfect vacation in open-ended natural language ("a peaceful place in nature, without crowds") → **Sentence-BERT model** converts the input into a vector representation that is then compared to **560 destination descriptions** → **Hybrid ranking mechanism** takes into account semantic similarity as well as flexible preferences (culture, nature, nightlife, etc.) → And users can tweak the weight of either **"what I wrote"** or **"what I chose using the sliders"** ⚙️ **Tech stack:** * **NLP:** Sentence Transformers (semantic embeddings) * **Backend:*