AI for Recruiting in 2026: Frappe HRMS (8k Stars) + LLM Resume Screening - Free ATS for Small Teams

Frappe HRMS (8,436 stars) is a free open-source HR suite with recruitment modules, and an LLM layer screens resumes against job requirements - a private ATS for small teams without per-seat SaaS fees.

💡 What You Will Learn

Frappe HRMS (8,436 stars) is a free open-source HR suite with recruitment modules, and an LLM layer screens resumes against job requirements - a private ATS for small teams without per-seat SaaS fees.

## The short answer **Frappe HRMS** (8,436 stars, GPL-3.0) covers the full hiring loop - job posts, applicant tracking, interviews, offers - for free. The AI add-on: extract resume text (PyMuPDF, 10,400 stars), score against the job description with an LLM, and rank candidates before a human ever reads them. ## The screening pipeline ```python import fitz, openai, json def score_resume(pdf_path, job_reqs): doc = fitz.open(pdf_path) text = "".join(p.get_text() for p in doc) resp = openai.chat.completions.create( model="gpt-4o-mini", messages=[{"role": "user", "content": f"Score this resume against: {job_reqs}. Return JSON: " "{"score": 0-100, "matches": [...], "gaps": [...], "notes": "..."} " + text}]) return json.loads(resp.choices[0].message.content) ``` Run it over all applicants, sort by score, and shortlist the top 10-20% for human review. ## Why self-host | Aspect | Free stack | Typical ATS SaaS | |:-------|:-----------|:-----------------| | Cost | $0 | $99-499/month | | Resume data | Your server | Vendor cloud | | Custom fields | Full control | Limited | | LLM screening | Any model | Vendor's model | ## Real numbers - Screening 200 resumes with an LLM costs ~$1-3 (mini-class model) and takes ~20 minutes. - A human screener averages 3-5 minutes per resume - 200 resumes is a full day. - Frappe HRMS includes payroll, leave and expense modules too - one system for the whole HR function. ## FAQ **Q: Does the LLM replace human screening?** A: No - it pre-ranks and flags gaps. Final shortlisting should always involve a human reading the top candidates. **Q: Is resume data safe?** A: Self-hosted Frappe + local LLM (Ollama, 177,825 stars) keeps all resumes on your infrastructure - important for GDPR-style requirements. **Q: What about LinkedIn integration?** A: Not built-in; export applications as files and let the pipeline ingest them.