AI Powered Recruitment Agent

The AI Powered Recruitment Agent automates the first round of hiring for small companies, where hundreds of resumes can arrive for one job and manual screening is slow and inconsistent. Recruiter side: Recruiters post jobs with required skills and experience. When applicants apply, the AI reads each resume, scores it 0-100 against that job, explains the score, and lists matching and missing skills. Recruiters see a ranked list with filters, then shortlist, hold or reject candidates. For shortlisted candidates the AI writes tailored technical and behavioral interview questions. A dashboard shows statistics, and ranked lists export to CSV or PDF. Applicant side: Applicants browse jobs and apply with a PDF or DOCX resume. They instantly see their match score, their skill gaps and AI feedback on how to improve the resume for that role. AI Tools: Anyone can also use the AI separately for resume screening, interview question generation, and resume feedback, without posting or applying. Technology: Python and Flask run the backend, with SQLite and SQLAlchemy for data, Bootstrap and JavaScript for the interface, and Chart.js for charts. Google Gemini is the main AI model, with Groq as a fallback and a rule-based matcher as a last resort. PyMuPDF and Python-docx read resumes, APScheduler and Gmail SMTP handle scheduled emails, and ReportLab builds PDF reports. Built with Antigravity and hosted on Render. Why not Vultr: The app is designed for a Vultr VM, but the hackathon credit pool ran out and Vultr required card verification before we could start. We raised this in a ticket on the lablab Discord, and the organizers told us to submit our repository, video and slides and to state this in the README and description. We deployed a temporary demo on Render instead. The README holds our Vultr deployment plan.