RAG demo · 100% local
Match job ↔ CV
Paste a job description below. The tool computes semantic similarity between the description and the projects, experiences and skills detailed on this site, then surfaces the strongest matches. Everything happens in your browser — no text is sent to any server.
Job description
Paste the description text, or upload a PDF. The embeddings model
(all-MiniLM-L6-v2, ~25 MB) loads once and is cached afterwards.
Ready.
Matches
Results ranked by similarity score will appear here after analysis.
Under the hood
The tool loads Transformers.js
(JavaScript port of Hugging Face Transformers) together with
Xenova/all-MiniLM-L6-v2, a sentence encoder trained to place
semantically similar texts close together in vector space.
A static corpus contains the descriptions of the projects, experiences and
skills presented on this site. On first load, each description is turned
into a 384-dimensional vector and cached in localStorage.
When you press Analyze, the job description is vectorised the same way, and cosine similarity is computed between its vector and each vector in the corpus. The five highest-scoring entries are displayed with a link to the corresponding page on the site.
Neither the job description nor the results ever leave your tab. No server is called at runtime — only the model and corpus file are downloaded once.