System Requirements Document for screening-web

1. Introduction

Product intent. screening-web is a full-stack website for AI resume screening and ranking. A Recruiter uploads multiple PDF resumes, enters the job description they are hiring against, and receives a ranked candidate list that shows each candidate's name, matched skills, missing skills, and an overall matching score. The product exists to turn a pile of unstructured resume PDFs plus a free-text job description into an ordered, comparable shortlist that a recruiter can act on directly from a clean, modern, responsive dashboard.

Audience. The primary human audience is the Recruiter / Hiring Reviewer running a screening batch and deciding who to advance. A second accepted human audience is the Developer / Maintainer who stands the project up from the README, runs the backend API and database, loads the sample dataset, and executes the unit tests. Both audiences are served by this system, but through different surfaces: the Recruiter through the browser application, the Developer / Maintainer through repository documentation, the sample dataset, the test suite, and the backend API.

Deliverable shape. The system ships as a full-stack web application (frontend, backend API, database), with input validation and error handling, a sample dataset for testing, unit tests for the main backend functionality, and a README covering architecture, setup instructions, API endpoints, and the testing process. Requirements, architecture, and implementation plan are documented and reviewed before development begins.

2. System Overview

screening-web is a browser application backed by a first-party API and database.

Actors.

• Recruiter / Hiring Reviewer (human, accepted): opens screening-web directly in the browser with no sign up and no login, creates screening batches by uploading multiple PDF resumes and entering a job description, and reviews the ranked candidate list with names, matched skills, missing skills, and overall matching scores.
• Developer / Maintainer (human, accepted): sets up the project from the README, runs the backend API and database, loads the sample dataset for testing, and executes the unit tests covering the main backend functionality.
• Screening backend (non-persona system actor): validates input, stores resumes, job descriptions, and results, extracts