Resume Roaster is a lightweight web application that humorously critiques resumes using rule-based pattern matching. I built it as a fun side project, which felt very relevant in a difficult job market where everyone is so focused on optimizing their resumes. It also served as a great way to practice my JavaScript and rapid web deployment.
The system is intentionally simple and transparent, favoring fast, explainable feedback over complex machine-learning models.
Live site: https://radkinz.com/resume-roaster/website/
- A user uploads their resume as a pdf.
- After converting the pdf into a massive string, which is decomposed by section. The JavaScript logic applies a set of handcrafted rules and patterns to identify common resume patterns.
- Each detected pattern maps to a list of various short, humorous critiques, and one is picked by random.
- Results are returned instantly with no external API calls or model inference.
The project deliberately avoids machine learning in favor of:
- Full control over outputs
- Predictable behavior
- Easy extensibility and debugging
- Fully client-side resume analysis (runs entirely in the browser)
- Rule-based text analysis using pattern matching
- Instant feedback with no server requests or data storage
- Lightweight UI
git clone https://github.com/radkinz/resume-roaster.git
cd resume-roaster
python3 -m http.server