03 / AI / applied tools
Matchya — CV screening
Matchya turns a pile of CVs into a ranked spreadsheet with reasons for each score. Built with Elizaveta Kalinina for a real screening task.
Demo walkthrough
- 01
Set a Python vacancy and weighted criteria
- 02
Upload and process three CVs
- 03
Export ranking, scores and reasons
Three candidates ranked with scores, priority groups and explanations. The workbook also includes criteria statistics and run settings.
Download demo ranking · Excel ↓The problem
We needed to screen around a thousand candidates within a few hours. The output had to be useful to the person making the next decision: contacts, comparable scores and an explanation.
What I built
- Scored CVs against weighted job criteria and included the reasoning behind each score.
- Extracted contacts and removed duplicates by file hash, email and phone number.
- Delivered a ranked Excel report for human review, with support for OpenAI-compatible model providers.
What changed
1,000 / <1 h
The reported run processed 1,000 CVs in under an hour. Candidates with the highest scores progressed to the next stage; the ranking supported human selection.
Source & scopeResult reported in our project portfolio. Throughput describes that run; it is not a general accuracy benchmark.
Tools
- Python
- LLM APIs
- vLLM
- Excel