AI Resume Judge

Problem it solves

Resume screening often collapses into subjective pattern matching. The same line about “K8s operations” may mean tooling execution or architecture ownership; “improved performance” may include latency, cost, and QPS, or it may be only boilerplate. AI-Resume-Judge turns that judgment into auditable dimensions and checks whether the text evidence shows complete contact information, quantified impact, layered technical depth, and career progression.

Workflow and limits

It accepts Markdown text or a file path, then runs AI field detection. If it finds strong signals such as LLM, RAG, Agent, or PyTorch, or enough weaker signals, it scores the resume against the AI benchmark. For non-AI roles, it maps the role to an industry coefficient and converts the base score, for example DevOps uses 0.72.

Scoring is on a 100-point scale:
- Basic information completeness (15): email, phone, GitHub, LinkedIn, location.
- Quantified achievements (25): numbers, percentages, business impact, cost/benefit.
- Technical depth and breadth (20): stack, LLM/Agent/RAG-level concepts, proficiency levels.
- Work experience quality (20): company signal, ownership versus participation, growth trajectory.
- Education, brand/influence, and expression make up the remaining 20 points.

Penalties apply for vague verbs such as “participated in”, missing metrics, flat skill lists, and boilerplate self-evaluation. It outputs a structured Markdown report and can use report_template.html for a visual version. Treat it as a text-level screening aid, not proof of interview performance; cross-industry conversion is only as reliable as the industry coefficient mapping.