AI CV Screener
MugDevs needed a faster way to get from 100+ applications to a shortlist worth reading. This workflow watches a Drive folder, processes every CV automatically, scores each candidate against four defined criteria using Gemini, and logs the results to a sheet the hiring team can act on immediately.
The problem
MugDevs posted a Senior Frontend Engineer role and had 70+ CVs sitting in a Drive folder within two weeks. The folder stayed untouched.
The real issue wasn't volume. It was that reading CVs is slow, inconsistent, and mind-numbing work. Two people reviewing the same stack of CVs will shortlist different people based on what they happened to notice. The process needed to be faster and more consistent.
What this workflow does
The workflow monitors a Google Drive folder. When a new CV lands, it gets picked up automatically.
PDFs go through text extraction. The extracted text gets sent to Gemini with the job criteria baked into the prompt: React and TypeScript experience, Next.js shipped in production, total years of frontend experience, and whether the candidate has worked at a product company. Gemini scores each criterion, returns a total score, and writes a short justification.
That data gets logged to Google Sheets. One row per candidate, with scores, evidence pulled from the CV, and a status column that tells the hiring team whether the candidate passed, failed, or needs a manual look.
Files that can't be read automatically, scanned PDFs, Word documents, get flagged separately so nothing falls through the cracks.
How it was built
The trigger watches the Drive folder for new files. A loop processes each CV one at a time. A router checks the file type and sends PDFs to the extraction node. The extracted text goes to Gemini via HTTP node with the scoring prompt. A code node parses the response into structured fields. Google Sheets logs everything.
DOC and DOCX files, along with image-based PDFs that can't be read as text, get routed to a separate branch and flagged for manual review with the filename preserved so nothing gets lost.
The result
The hiring team opens one spreadsheet. Every readable CV has a score, a justification, and a pass or fail status. Files that need a human look are flagged clearly. Nobody has to open a single CV to build the initial shortlist.
Got a folder of CVs nobody has time to read?
Tell me the role, the criteria, and where your applications land. I'll build something that shortlists them before your hiring manager finishes their morning coffee.
More projects
AI Social Media Repurposer
Paste a blog URL, get a LinkedIn post, a tweet, and an email snippet back. All three land in a Google Sheet with the source URL and timestamp.
AI Email Response Generator
Reads every support email, classifies it, drafts a context-aware reply, and saves it as a Gmail draft ready to send. The team reviews, clicks send, and moves on.