Job applicant CV scored by AI and logged to Google Sheets

Job applicant CV scored by AI and logged to Google Sheets

An n8n workflow with a CV form. An applicant sends a name, an email address and a PDF. The workflow keeps the PDF in
Google Drive, has an OpenAI model pull out the contact details and the career, summarizes them, and scores the applicant
from 1 to 10 against a profile you wrote. One row per applicant lands in a Google Sheet.

What is inside

• workflow.json: the workflow as n8n exports it, 12 nodes beside 6 sticky notes. It holds no credential.
• README.md: these notes.

What we changed from the original copy

The copy Mindket started from had two slips that left a column empty. Mindket fixed both:

• Google Sheets read the applicant's name from Nome, but the form's field is called Name, so the NAME column
  stayed blank. It now reads Name.
• The same node filled a column called TELEFONO, while the sheet's column list in the file calls it PHONE. A row is
  written by header name, so the phone number never arrived. The value now goes to PHONE.

The sheet's address and the Drive folder's address belonged to the original copy's own account. They now read
YOUR_SHEET_ID and YOUR_FOLDER_ID, and the cached names of that sheet are gone.

Before you start

• An n8n instance, on n8n Cloud or your own server.
• An OpenAI API key. The one chat model node uses gpt-4o-mini.
• A Google account that can use both Drive and Sheets.
• A Drive folder for the CVs.
• A Google Sheet whose first row holds these twelve headers, spelled exactly as written: DATA, NAME, PHONE,
  CITY, EMAIL, DATA NASCITA, EDUCATIONAL, JOB HISTORY, SKILLS, SUMMARIZE, VOTE and CONSIDERATION. The
  node stops with "Column names were updated after the node's setup" when one is missing. Extra columns are fine.

Set it up

1. In n8n, create a workflow, open its menu and choose Import from File. Pick workflow.json.
2. Add an OpenAI credential and select it in OpenAI. That one node serves all four AI steps.
3. Add a Google Drive OAuth2 credential and select it in Upload CV. Replace YOUR_FOLDER_ID with your folder's ID,
   the part of its address after /folders/.
4. Add a Google Sheets OAuth2 credential and select it in Google Sheets. Replace YOUR_SHEET_ID with the sheet's ID,
   the part of its address between /d/ and /edit. The node writes to the sheet's first tab.
5. Open Profile Wanted and replace its text with the profile you are hiring for.
6. Save, open the form's test address, and send a test CV of your own as a PDF.
7. Check the new row, then press Publish and share the form's production address.

How a run goes

1. On form submission is the form, titled Send CV. It takes Name, Email and CV, which must be a .pdf.
2. Upload CV saves the PDF in your Drive folder as CV- plus the date and the original file name.
3. Extract from File reads the text out of the PDF.
4. Personal Data asks the model for the telephone number, the city and the birthdate. Qualifications asks for the
   education and the job history, each in 100 words at most, and for the skills as a bulleted list.
5. Merge joins the two answers into one item.
6. Summarization Chain writes a summary of 100 words or less, in a conversational tone.
7. Profile Wanted holds the profile text. It ships with a web agency looking for a full-stack developer in northern
   Italy.
8. HR Expert compares the summary with that profile. A Structured Output Parser makes it answer with a vote and a
   consideration.
9. Google Sheets appends one row: the date, the name, the phone, the city, the email, the birthdate, the education, the
   job history, the skills, the summary, the vote and its reasons.

Change the look

• The profile: the text in Profile Wanted.
• The scoring rule: the message in HR Expert. It asks for a vote, where the top mark is the ideal candidate, and
  for the reasons in consideration. It is written in Italian, as are the labels Profilo ricercato and Candidato.
  Translate it if you want the reasons in another language.
• What is pulled out: the attributes of Qualifications and the schema of Personal Data.
• The summary: the prompt in Summarization Chain.
• The columns: the Google Sheets node and the sheet's first row, which must stay in step.
• The model: OpenAI.

Good to know

• A person should read the vote. It is a model's opinion of a summary of a PDF, and a badly laid out CV can lose
  details before the model sees them.
• Each run sends the CV's text to OpenAI and writes the applicant's phone number and birthdate to your sheet. Tell
  applicants what happens to their data, and check the hiring and data-protection rules where you operate.
• Upload CV is a branch of its own. Nothing later reads the saved file, so the sheet holds no link to it.
• The form takes PDFs only, and a scanned PDF with no text layer gives the model nothing to read.
• OpenAI's deprecations page, read on 12 November 2026, gives 14 January 2027 as the shutdown date of gpt-4o-mini
  and names gpt-6-luna as its replacement. Pick the replacement in OpenAI before then.
• Billing: OpenAI bills at least four model calls for each CV, one in each of Personal Data, Qualifications,
  Summarization Chain and HR Expert.
• Once the workflow is published, anyone with the form's address can send CVs and spend your OpenAI key.
• The JSON holds no credential. The Profile Wanted text is an example for the original copy's own search.