I want to build an AI-driven recruitment assistant focused solely on resume screening. The system must ingest and parse both text-based and PDF resumes, then rank or flag candidates against configurable criteria such as keywords, years of experience, qualifications, and any red-flag gaps. Because I plan to use it as a self-contained solution, please design it as a standalone web app (a lightweight dashboard or simple API is fine). There is no need to plug into existing HR suites, but export options—CSV download and a basic JSON endpoint—will help me move data wherever I need. Key points I care about • Accurate parsing of text and PDF formats, including multi-column layouts. • Transparent scoring logic so I can tweak weightings without touching the code. • A small sample dataset or clear instructions for training on my own corpus. • Clean UI for uploading files in bulk and viewing ranked results. I’m comfortable with common NLP stacks—Python, spaCy, scikit-learn, or a compact Transformer model—so use whichever combination delivers speed and clarity. As long as the code is well-commented and supplied in a Git repo, I’ll handle hosting. Acceptance criteria 1. Upload 100 mixed resumes (PDF + TXT) and receive a ranked shortlist in under one minute. 2. Editable YAML/JSON config lets me modify keyword lists and weightings. 3. Model performance summary (precision/recall) provided on the sample set you supply. 4. README with setup, run, and retraining steps straightforward enough for a non-ML engineer to follow. Once delivered and tested on my side, I will sign off promptly and discuss future enhancements like interview scheduling or ATS integration.