I built an AI tool to rank 200+ CVs in seconds (FastAPI + embeddings)
Small businesses post a job, get 200+ CVs and spend days reading them manually.
Most end up skimming the first 50 and hoping for the best.
I kept seeing this problem, so I built a simple tool to fix it.
What it does
Upload a batch of CVs (PDF or DOCX), describe the role in plain English, and get a ranked shortlist in seconds.
No filters, no complex setup.
Just:
- Upload CVs
- Paste job description
- Get ranked candidates
Why I built it
I’ve worked with small teams and founders who don’t have HR departments.
They don’t need another ATS.
They just need:
- a quick way to not miss good candidates
- a way to avoid reading 200 PDFs manually
How it works
Each CV is converted into an embedding.
The job description is also converted into an embedding.
Then I calculate similarity between them and rank candidates.
No keyword matching. No rules engine.
Tech stack
- FastAPI
- SQLite
- Hugging Face Inference API (BGE embeddings)
- Tailwind CSS
- Hosted on Render
Code is open here:
👉 GitHub repo
Demo
Things that surprised me
- Embeddings outperform keyword matching by a lot
- CV parsing is messy (PDFs are chaos)
- Users don’t want features — they want results fast
What I’m exploring next
- Handling very large batches efficiently
- Adding explainability (why a CV ranks higher)
- Whether recruiters would use something this simple
Try it
Would love feedback from developers or anyone involved in hiring.
Top comments (0)