Resume Editor

npx latex-resume-cli

I got tired of manually tweaking my resume for every job application. So I built this.

It's a resume editor that runs on AWS. Tell it what you want in plain English ("add 2 years at Google doing Kubernetes stuff") and an AI generates the LaTeX for you. The whole thing costs $0 when you're not using it.

Quick Start

You: "Add 2 years at Google doing Kubernetes stuff"
     ↓
AI generates LaTeX patch → compiles PDF → you preview it
     ↓
Accept → auto-commits to GitHub

The backend spins up when you need it, shuts down after 10 minutes of inactivity. No always-on servers burning money.

What's Inside

Everything runs in one Docker container on ECS Fargate Spot:

• Bun - handles API requests, serves the frontend
• Next.js - static export, bundled into the container
• TeX Live - pdflatex + latexmk for compilation
• Cloudflare Tunnel - exposes the service without paying for a load balancer

Why Fargate Spot?

Lambda would've been the obvious choice, but TeX Live is huge (~2GB). Fitting it into a Lambda layer is a pain. With Docker I just apt-get install what I need and move on.

Fargate Spot is ~70% cheaper than regular Fargate. I only run for a few minutes at a time, so Spot interruptions are a non-issue.

Why Bedrock?

No API keys to rotate. Bedrock runs in my VPC, auth is handled by IAM. Qwen on Bedrock is pay-per-token - way cheaper than a ChatGPT subscription for occasional resume tweaks.

Wake-on-Demand Flow

1. Idle - Nothing running. Cost: $0
2. You visit the site - Lambda wakes up ECS (~45-60s cold start)
3. You edit - Backend is live
4. 10 min idle - Container kills itself. Back to $0

Budget Kill Switch

There's a failsafe: if AWS costs hit 80% of your monthly budget ($4 out of $5 by default), a Lambda automatically:

1. Stops all ECS tasks
2. Disables the wakeup Lambda (sets concurrency to 0)

This prevents runaway costs. To re-enable, just reset the Lambda concurrency in the AWS console.

Setup

You'll Need

• AWS account with admin access
• Cloudflare account (free tier works)
• Terraform installed

Deploy

1. Fork/clone this repo

2. Add GitHub secrets:

   • AWS_ACCESS_KEY_ID
   • AWS_SECRET_ACCESS_KEY

3. Configure Terraform:

   cd terraform
   cp terraform.tfvars.example terraform.tfvars
   # edit with your values

4. Deploy:

   terraform init
   terraform apply

5. Push to main - GitHub Actions builds the image, pushes to ECR

terraform.tfvars

github_token            = "ghp_..."           # for committing resume changes
repo_owner              = "your-username"
repo_name               = "your-repo"
cloudflare_tunnel_token = "eyJ..."            # from Cloudflare Zero Trust
budget_alert_email      = "you@example.com"   # budget notification email

Using It

1. Run npx latex-resume-cli (or go to your Cloudflare tunnel URL)
2. Wait ~60s for Fargate to spin up
3. Type what you want changed
4. Preview the PDF
5. Accept → commits to GitHub

Project Layout

├── compute/          # Bun backend
├── web/              # Next.js frontend (static export)
├── terraform/        # All the infrastructure
├── lambda_src/       # Wake up, stop, and budget kill Lambdas
├── cli/              # The npx command
├── resume.tex        # Your resume
└── Dockerfile        # Everything bundled

---

Built by Vishal Shaji