Talenta (ATS Analyzer)
AI-powered CV scoring tool for job seekers: 0–100 compatibility score, missing keywords, section analysis, and rewrite suggestions.
« Remplacer "Développement d'APIs" par "Conception d'APIs REST sous Node.js avec réduction de latence de 35%" pour correspondre au critère de performance requis. »
The Vision
Empowering job seekers to beat automated ATS filters through transparent AI-driven resume scoring and actionable keyword optimization.
Why this project?
Over 75% of resumes are discarded by automated Applicant Tracking Systems before a human recruiter ever sees them due to formatting glitches or missing keywords.
Technical approach
Built a full-stack Nuxt 4 web app utilizing Nuxt UI v4 (Radix Vue + Tailwind), Supabase for auth, PostgreSQL with Row-Level Security, and Storage. AI inference queries HuggingFace (Mistral-7B) with structured Zod schema parsing. Integrated Stripe webhooks for subscription quotas (Free, Pro, Expert) and Resend for transactional email dispatch.
Expected outcome
Immediate 0–100 score feedback, ranked matched/missing keywords, section-by-section breakdown (skills, experience, education), and actionable before/after sentence rewrites.
Dossier d'Architecture & Spécifications
Conception logicielle, flux de données, sécurité SQL et contrats d'API
01 / Use Cases & Business Rules
Parsing multipart de CV (PDF/DOCX), assainissement Zod et matching avec la fiche de poste.
Détection des hard-skills manquants et reformulation de phrases selon la méthode X-Y-Z.
Décrémentation atomique par fonction stockée PostgreSQL et webhooks Stripe Checkout.