Screen and manage candidates with Gmail, Google Sheets, Google Drive and Nvidia Llama

Quick overview

This workflow ingests PDF resumes from Gmail (including Indeed application emails),extracts candidate details with an NVIDIA NIM LLM, stores the resume in Google Drive, and appends a deduplicated candidate record to Google Sheets, then sends acceptance or rejection emails and updates the sheet based on recruiter decisions and candidate replies.

How it works

1. Triggers every minute when a Gmail message arrives with a PDF attachment (or runs from a manual local-PDF test path) and normalizes the resume file plus sender and role metadata.
2. Extracts text from the PDF and sends it to the NVIDIA NIM chat completions API to parse structured candidate fields (name, email, phone, LinkedIn, portfolio, previous company, and position).
3. Uploads the original resume PDF to Google Drive and combines the Drive file link with the AI-parsed candidate data.
4. Checks the Google Sheets “Candidates” tab for existing emails to deduplicate, formats a candidate row (including a Drive hyperlink), and appends the new record to Google Sheets.
5. Triggers on a webhook or runs every minute to read Google Sheets decisions, filters rows marked Approve/Reject where no email has been sent, and routes them by decision.
6. Sends an acceptance questionnaire or a rejection email via Gmail and updates the corresponding Google Sheets row with the email status and processing timestamp.
7. Monitors Gmail for candidate replies to the questionnaire, uses the NVIDIA NIM API to extract questionnaire answers (CTC, notice period, location, etc.), and updates the matching Google Sheets row by candidate email.

Setup

1. Connect Gmail OAuth2 credentials for both reading incoming emails (resumes and replies) and sending acceptance/rejection emails.
2. Connect Google Sheets OAuth2 credentials, create a spreadsheet with a sheet named “Candidates”, and replace all instances of YOUR_SPREADSHEET_ID in the workflow (including the Sheets API call in the formatting step).
3. Connect Google Drive OAuth2 credentials and set YOUR_GOOGLE_DRIVE_FOLDER_ID in the Google Drive upload step.
4. Add your NVIDIA NIM API key by replacing the Bearer YOUR_API_KEY headers in both AI HTTP request steps.
5. If you use the webhook approval path, copy the production webhook URL for /candidate-approval and configure your approval source to POST the row data (including Email, Candidate ID, human_approval, and Email Status).