AI Resume Screening System
Resume lands in Gmail. Ninety seconds later the hiring sheet has strengths, weaknesses, risk, reward, and a 0 to 10 fit score.
The problem
"Manually reading every resume against the same job description was burning the hiring manager's week. We needed a first-pass screen that was consistent, ranked, and left an audit trail."
The solution
One flow. Every input type.
The Gmail trigger polls every minute and downloads any attachments. A switch node routes on MIME type: Word, PDF, or plain text. Word docs are converted to Google Docs via a raw HTTP call to the Drive API (?convert=true), then exported back to PDF for a clean text extract; PDFs and text files go through their own extract paths. All three routes converge on a Set node that stores the resume text under a single key. The workflow then downloads the job description from a fixed Google Doc, extracts its text, and hands both to a LangChain recruiter agent running o4-mini. A structured output parser enforces the exact JSON shape: strengths and weaknesses as arrays, risk and reward each as a score plus explanation, an integer 0 to 10 fit rating, and a written justification. A second information extractor pulls first name, last name, and email from the resume. The full record appends to a Resume Screener Google Sheet with the Drive link to the original file.
Results
Industry
Recruiting / HR Ops
Service type
AI Automation
Timeline
1 week
Tools used