Resume Parsing & AI Candidate Screening

This is a proper Frappe app — its own module, DocTypes, hooks.py doc events, and server scripts — installed onto your ERPNext v15 or v16 bench, not a bolt-on SaaS. When a Job Applicant is created (via the careers webform, an email inbox, or the REST API), a before_save/after_insert doc-event hook enqueues a background job that pulls the attached resume, extracts text from PDF and DOCX, and parses it into structured child-table fields: contact details, work history, education, certifications, and a normalized skills list. We add a Resume Profile DocType linked one-to-one to Job Applicant so the parsed data is queryable, reportable, and permission-controlled like any native record. Skills are extracted and written to a tagged child table so you can filter and report on them across every opening.

The AI match engine reads the linked Job Opening — its description, required skills, and experience thresholds — and computes a 0–100 match score per candidate, with a written rationale stored on the record so the decision is auditable, not a black box. A whitelisted method (@frappe.whitelist()) exposes on-demand rescoring, and a scheduler event can batch-score any backlog on the nightly cron. Results drive an auto-shortlist: candidates above your configurable threshold are advanced and tagged, and (optionally, behind a manual approve step) templated rejection or acknowledgement emails fire through Frappe's Email Queue. Bias-aware controls let you blind configured fields (name, gender, age, photo, nationality) from the scoring prompt and log every automated decision for compliance review. Everything respects ERPNext permissions and role profiles — a recruiter sees their reqs, a hiring manager sees the shortlist, and raw PII stays gated.

Because this is build-to-order, we start from your actual hiring workflow, not a generic demo. After a short scoping call we confirm the DocType design, which resume fields matter to you, your scoring rubric, and your bias-blinding and email policies, then build against a staging bench. You get the installable app source for your version, UAT on staging with a rollback plan, and a git repository handover so you own the code outright. Typical delivery is 2–4 weeks from confirmed scope. There is no instant download — this is engineered, tested on a copy of your data, and installed by ECOSIRE.