Overview
This project focuses on building a practical evaluator that outputs actionable, structured results rather than vague AI feedback. It’s designed to be extended into a web tool or internal HR assistant.
LLMStructured OutputEvaluationPrompting
AI / LLM Project
CV Evaluator LLM
An LLM pipeline that evaluates resumes against a role, generates structured feedback, surfaces missing signals, and suggests improvements with consistent scoring.
GitHub RepoResume evaluation
Key outputs
- Fit score and rationale aligned to the role.
- Strengths and gaps (missing signals).
- Recommendations to improve the CV for the target job.
- Structured JSON (easy to integrate into apps).
How it works
- Input: CV text + job description (or role requirements).
- Normalization: extract key sections and entities such as skills, experience, and education.
- Evaluation: criteria-based scoring with calibrated explanations.
- Output: formatted report + JSON schema for automation.
Next improvements
- Add rubric tuning per industry (SWE, Data, Product, etc.).
- Bias and safety layer: avoid sensitive inference and focus on job-relevant content.
- Optional: RAG with role-specific standards and competency frameworks.
- Deploy a minimal web demo with Streamlit or FastAPI.