ReTeX

• 

  Logo

Inspiration

The modern job market is a numbers game. As students seeking co-op jobs, we realized that the most time consuming part of the application process is manually tailoring resumes for every unique job description. We built ReTeX to eliminate this friction by providing a tool that increases productivity and ensures candidates' resumes and cover letters are engineered for every application.

What it does

ReTeX processes a user's resume and a target job description to generate a complete application package. The system intelligently maps skills and experiences to output a refined Resume and a tailored Cover Letter. Beyond just writing, it provides a Strategic Advice section which personalized recommendation box that suggests specific actions the user can take to better match the job's requirements. To ensure maximum flexibility, the tool provides both high resolution PDF previews and the raw LaTeX source code via a direct integration with Overleaf for final manual editing.

How we built it

The technical foundation is a Spring Boot 3 backend leveraging Java 17. We utilized Spring AI to orchestrate requests to the Google Gemini 2.5 Flash model. To handle unstructured data, we integrated Apache Tika and Apache POI, allowing the system to process content from various formats including PDFs and Word documents. The frontend is built for a seamless, single page experience using HTMX, which allows for dynamic result rendering on localhost without full-page refreshes. The architecture follows a strict MVC pattern and uses environment variables to ensure secure API credential management.

Challenges we ran into

Development was a journey of technical troubleshooting and security hardening. We initially faced a setback with revoked API keys due to public exposure, which led us to reengineer our security protocols using .env files. We also grappled with the messiness of document extraction by transforming raw, unstructured PDF text into a formatted PDF and readable LaTeX file. Finally, ensuring the AI’s generative response could be reliably converted into valid LaTeX syntax without breaking the PDF compiler required complex processing and prompt engineering.

Accomplishments that we're proud of

We successfully engineered a system that extracts text from multiple file types and converts raw AI output into a professional, typeset document. We are particularly proud of our Overleaf Integration, which bridges the gap between AI generation and user controlled editing. Building the Tone Selection logic which allows user to pivot their application's voice for different personal preferences.

What we learned

This project served as a deep dive into AI orchestration and software security. We mastered prompt engineering to ensure Gemini provides factual, professional results and gained a deep respect for environment variable security. Beyond the AI, we expanded our knowledge of the Java ecosystem, specifically using Tika and Jackson to solve real-world data processing problems and learning how to integrate complex third-party tools like the Overleaf editor into a local development environment.

What's next for ReTeX

The future of ReTeX involves scaling from a local tool into a Multi-Model Career Suite. We envision a platform that aggregates insights from several AI models (OpenAI, Claude, and Gemini) to provide the most objective and balanced career advice possible. We plan to introduce premium features such as 1-on-1 AI-simulated interviews and automated, deep dive resume audits, transforming ReTeX into a comprehensive, personal career coach for all job seeker.

Built With

• gemini
• java
• maven
• overleaf
• springai
• springboot