Semantic Word Suggestion Search Box

I need a lightweight web component that behaves much like Google’s search bar: as soon as a user types a word, the interface should instantly display semantically related terms beneath the field. The aesthetic should stay strictly minimalist—clean edges, no visual distractions, quick to load.

Core requirements
• Real-time lookup of the nearest neighbors pulled from pre-trained word embeddings. I am able to learn GloVe, Word2Vec, fastText or other well-documented models if you can justify the choice and explain usage and or plan.

• Out-of-the-box English support, but please architect the code so additional languages can be swapped in by simply loading a different vector file.

• Response times under a few hundred milliseconds for typical queries on a modest server.
• Clear instructions on where the model files live, how they are loaded, and how the cosine-similarity (or similar) lookup works.

ability to integrate on current live website
Deliverables
- Front-end HTML/CSS/JS (or a small React/Vue component) that renders the minimalist search box and live results list.
- Back-end script or API endpoint (Python/Flask, Node, etc.) that performs the nearest-vector search.
- README that shows setup, vector download links, and how to extend to new languages.
- Brief test dataset or unit test proving the suggestions work.

Acceptance criteria
A short demo GIF or hosted link where I can type “happy” and immediately see synonyms such as “joyful”, “cheerful”, and “content” confirms completion.

Keep the code clear, dependency-light, and ready to drop into an existing site.