AI Agent for ATS‑Friendly Resume Optimization
Generate tailored, ATS-ready LaTeX/PDF resumes from a JD
MVP tool that ingests a job description (URL or raw text) and rewrites resume sections to improve ATS compatibility. Outputs LaTeX/PDF and a JSON report detailing coverage and changes. Provides a Streamlit UI for uploads and advanced options, with diagnostics for LaTeX compilation to simplify troubleshooting.
Timeline
1 week
Role
Full-Stack Developer
The Problem
Manual tailoring of resumes to specific job descriptions is time-consuming and error‑prone, and often fails automated ATS screening.
The Solution
Automate keyword extraction and bullet rewriting using LLMs, then render a polished LaTeX resume and a transparent report. Offer both a scriptable CLI and a simple Streamlit UI for non-technical usage.
Key Features
Explore the main capabilities and functionality of this project
JD by URL or Text
Accepts job description via URL or raw text input
Optimization Strategies
Choose conservative, balanced, or bold rewriting approaches
Pages Policy
Target 1 page, 2 pages, or auto based on content density
Transparent Report
Outputs report.json with coverage and change details
PDF Generation
Builds resume.pdf using LaTeX; includes diagnostics for failures
Technical Challenges
Key challenges faced during development and how they were solved
Reliable LaTeX Compilation Across Environments
Users may lack full TeX distributions and run into missing packages or PATH issues.
Solution
Expose diagnostics/logging, document TEXBIN on macOS, and guide users to install minimal packages with tlmgr as needed.
Model Selection and Fallbacks
Different users prefer OpenAI or Gemini models; calls can fail due to rate limits or configuration.
Solution
Support both providers via env vars with an overridable --model flag and a fallback model for resilience.
Preserving User Formatting While Optimizing Content
Editing LaTeX while maintaining structure and style can be brittle.
Solution
Allow users to supply a TeX template and keep artifacts next to outputs; focus edits on content blocks rather than global style.
Results & Impact
Measurable outcomes and achievements from this project
OpenAI, Google
Supported Providers
20,000 chars
Max JD Length (text)
PDF ≤ 15 MB
UI Upload Limit
Key Achievements
Unified CLI and Streamlit UI with identical optimization capabilities
Robust LaTeX troubleshooting flow via diagnostics tab
Flexible template option to preserve personal formatting
Project Impact
Reduces manual tailoring effort and improves ATS pass likelihood by aligning resume phrasing and keywords with the target JD.
Technology Stack
Technologies and tools used to build this project
frontend
backend
database
tools
apis
Lessons Learned
Clear diagnostics greatly reduce support load for LaTeX/PDF issues
Small wording changes can materially improve ATS alignment
Provider-agnostic LLM layer increases reliability and user choice
Future Improvements
Inline PDF diff previews of changed bullets
Multi‑resume workspace with versioning
Job board integrations and auto‑fill of JD text
Interested in Learning More?
I'd love to discuss this project in detail and share insights about the development process.