Repository navigation
Shortlist v1.0.0
Production release of Shortlist, an AI-assisted resume screening and job recommendation web application built with Python, Flask, and scikit-learn.
Highlights & Implemented Features
-
Multi-Format Resume Ingestion: Parse text directly from PDF documents (
pypdf), Markdown (.md), plain text (.txt), or via direct text paste. - Candidate Profile & Contact Parsing: Automatic extraction of email addresses, phone numbers, candidate name heuristic, and total years of project experience.
-
Regex Boundary-Isolated Skill Extraction: Accurate detection across 100+ technical skills and aliases from
data/skills.jsonwithout false positives (e.g. "Java" vs. "JavaScript"). -
Targeted Job Screening Engine: Evaluates candidate fit against specific job descriptions using composite scoring (65% required skill coverage + 35% sublinear TF-IDF cosine similarity) with automated verdicts: Shortlist (
$\ge 70%$ ), Review ($45-69%$ ), or Not a fit ($< 45%$ ). -
Job Catalog Recommendations: Scores candidate profiles against an internal catalog (
data/jobs.json) using 55% required skill match, 15% nice-to-have skill match, and 30% text similarity, returning the top 8 matching roles. - Skill Gap Detection: Highlights the most frequent missing skills across top matching roles to guide candidates on upskilling.
- Responsive Single-Page Application: Interactive web UI with drag-and-drop file upload, radial match gauges, skill badge breakdowns, and sample data preloading.
- Privacy First & In-Memory: All resume parsing and scoring executes in-memory. No candidate data is persisted or sent to external cloud APIs.
Quick Start
- Download and extract
shortlist-v1.0.0.zip. - Create and activate a Python virtual environment:
python -m venv venv source venv/bin/activate # Windows: .\venv\Scripts\activate
- Install dependencies:
pip install -r requirements.txt
- Start the application:
python app.py
- Open your browser at
http://127.0.0.1:5000.
Checksums (SHA-256)
9e66adfd30b0cc3a3aea6cc8b45374937493d5fd4e6770fd17ebf7456cf4f178 shortlist-v1.0.0.zip
Known Limitations
- Scanned PDFs lacking an embedded text layer require OCR or manual text pasting.
- File uploads are capped at 5 MB and 15 pages.
Full Changelog: https://github.com/nfsprogramming/Shortlist/commits/v1.0.0
Full Changelog: https://github.com/nfsprogramming/Shortlist/commits/v1.0.0
Full Changelog: https://github.com/nfsprogramming/Shortlist/commits/v1.0.0