Inspiration

Job seekers spend hours tweaking their résumés for each opening, yet about 75 % are filtered out by Applicant Tracking Systems (ATS) before a human ever reads them. After helping friends—and myself—struggle with this pain point, I set out to automate résumé tailoring with AI so candidates can focus on interviewing, not formatting.

What We Learned

  • Semantic analysis beats simple keyword matching. Using embedding-based similarity models greatly improves skill extraction from job posts.
  • Prompt design matters: temperature, top-p, and system instructions drastically change ATS-ready keyword density.
  • Continuous A/B testing with real job applications shortens feedback loops and guides feature prioritization.
  • Privacy-first architecture (client-side parsing where possible) is essential for user trust.

How We Built It

Layer Technology Purpose
Frontend React (Vite) + Tailwind CSS Single-page app with drag-and-drop résumé upload
Backend Node.js (Express) + OpenAI/Claude APIs NLP, skill extraction, résumé regeneration
ATS Scoring Elasticsearch (BM25) Ranks how closely the tailored résumé matches the JD
Storage Supabase Postgres Secure storage for user docs and analytics
CI/CD GitHub Actions → Vercel Zero-downtime deployments on every merge

Pipeline

  1. Ingest job description → tokenize & embed
  2. Extract required skills via domain-tuned LLM prompt
  3. Compare to user résumé; identify gaps & synonyms
  4. Regenerate résumé sections with contextual prompts
  5. Score and preview: show ATS score before download

Challenges & Solutions

Challenge Attempted Fixes Final Solution
LLM hallucinations Prompt tweaks Rule-based post-filters + fact-check against source résumé
Keeping PDF under 200 KB Compression libraries Streamlined fonts and vector graphics during export
GDPR-grade data privacy Manual cleanup End-to-end encryption + auto-delete after 24 h
Cold-start latency on free tier Adding compute Edge caching for embeddings and memoization

Presicio Resume now lets candidates tailor a résumé in under 60 seconds, boosting interview rates while slashing prep time. Next milestones include multilingual support and one-click LinkedIn optimization.

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Precisio Resume

Built With

Share this project:

Updates