Resume Parser (Go + Groq LLM)

A command-line tool that lets you pick a resume file (PDF or DOCX) via a native file dialog, extracts its raw text, and uses an LLM (via the Groq API) to parse it into structured JSON matching a defined schema.

How it works

1. Pick a file — opens a native OS file browser (via zenity) restricted to .pdf and .docx files.
2. Extract text — reads the raw text content from the selected file:
   • PDFs are parsed with github.com/ledongthuc/pdf
   • DOCX files are parsed with github.com/lu4p/cat
3. Parse with LLM — sends the raw resume text to Groq's chat completions API (openai/gpt-oss-20b model) with a system prompt instructing it to extract fields matching a fixed JSON schema, and requests a json_object response format.
4. Structure the result — unmarshals the LLM's JSON output into a Go Resume struct and prints it.

Schema

The tool currently extracts:

type Resume struct {
    Name      string   `json:"name"`
    Email     string   `json:"email"`
    Phone     string   `json:"phone"`
    Skills    []string `json:"skills"`
    Education string   `json:"education"`
}

To parse against your own criteria, edit this struct and update the JSON schema described in the systemPrompt inside parseResumeWithLLM to match — the two need to stay in sync.

Prerequisites

• Go installed (1.20+ recommended)
• A Groq API key
• zenity installed on your system (for the file picker dialog):
  • Fedora: sudo dnf install zenity
  • Debian/Ubuntu: sudo apt install zenity
  • macOS: file dialogs work via a different native backend; no zenity install needed
• Windows: uses the native Windows file dialog; no extra install needed

Setup

1. Clone or copy this project into a folder.
2. Initialize the Go module (if not already done):

   go mod init resumeparser

3. Install dependencies:

   go get github.com/joho/godotenv
   go get github.com/ledongthuc/pdf
   go get github.com/lu4p/cat
   go get github.com/ncruces/zenity
   go mod tidy

4. Create a .env file in the project root:

   GROQ_API_KEY=your_groq_api_key_here
   

Usage

go run main.go

A file picker window will open. Select a .pdf or .docx resume. The tool will:

• Print the selected file path
• Extract and send the resume text to the LLM
• Print the parsed, structured fields

Project structure

Everything currently lives in a single main.go file:

Function	Responsibility
extractText	Reads a PDF or DOCX file and returns its raw text
pickResumeFile	Opens a native file picker dialog and returns the chosen path
parseResumeWithLLM	Sends resume text to Groq and returns a structured Resume
main	Wires the above together end-to-end

Known limitations / next steps

• Only .pdf and .docx are supported — no .doc, .rtf, or image-based/scanned PDFs (no OCR).
• The schema is currently hardcoded for demonstration; swap in your actual parsing criteria.
• No retry logic if the LLM returns malformed JSON — errors are surfaced with the raw response for debugging.
• No batch mode yet — one resume per run.

Troubleshooting

• no required module provides package ... — you forgot to go get that dependency. Run go mod tidy.
• open <file>: no such file or directory — the file picker didn't return the path you expected, or you hardcoded a filename that doesn't exist in your project folder.
• zenity: command not found / dialog doesn't open — install zenity for your OS (see Prerequisites).
• LLM output doesn't match the schema — check the raw response printed in the error message; you may need to tighten the system prompt or reduce resume text length.