JobTailor

Tailor any resume

DEMO VIDEO: https://www.loom.com/share/1c5dd698fc3f42da82707549c94308c8

Inspiration

We spend nearly two hours every week customizing our resume for job applications. It's one of the most effective ways to land interviews, but the process is repetitive and tedious — especially when dealing with LaTeX. With the Model Context Protocol (MCP), we realized we could automate the boring parts while keeping full control over the resume’s structure and quality.

What it does

JobTailor lets you edit your LaTeX resume using natural language. You describe the changes (e.g., “add a bullet under Databricks about my GraphQL security work”), and the system updates the .tex file accordingly. It then compiles the resume and shows you a real-time preview. The goal is to help you customize your resume for each role, fast.

How we built it

We used Claude AI as the front-end assistant, connected via MCP to a local server that defines functions like edit_text, add_section, and remove_item. Claude calls these functions to update the LaTeX source. The backend compiles the updated .tex file using pdflatex, and returns the PDF preview.

Challenges we ran into

We weren’t able to deploy the working version to Vercel in time due to the short hackathon window and build errors related to LaTeX compilation. Debugging interactions between Claude and the MCP function server also took longer than expected.

Accomplishments that we're proud of

We got the whole system working end-to-end: natural language → structured function calls → .tex edits → live PDF output. The edits worked cleanly, and Claude was able to understand and execute resume tweaks with minimal prompting.

What we learned

We learned how to build and run an MCP server from scratch and how to define callable functions for language models. None of us had worked with MCP before this project, so getting it working with Claude and seeing it successfully change LaTeX files was a big win.

What's next for JobTailor

We’re planning to rewrite and clean up the codebase, add support for common resume templates, and fully deploy the system on Vercel with persistent file storage and previewing. We genuinely think this tool can save hours for job applicants — and we’re serious about continuing to build it.

DEMO: https://www.loom.com/share/1c5dd698fc3f42da82707549c94308c8?sid=99bf1c2a-fe48-4709-b019-8e14896fe06b