AI hiring tools are powerful but carry significant legal and ethical risks. Technical implementation: 1) Resume parsing: extract structured information (skills, experience, education) using NLP (spaCy NER + rule-based). 2) Skill matching: compare extracted skills against job requirements using semantic similarity (skill embedding similarity + keyword matching). 3) Ranking: linear model combining skill match score, experience relevance, and qualification completeness. Legal considerations: Title VII (Civil Rights Act), ADEA (age discrimination), ADA (disability) apply to algorithmic hiring. NYC Local Law 144 requires bias audits for AI hiring tools. EU AI Act classifies high-risk. Bias mitigation: 1) Remove protected attributes and proxies from features (zip code can proxy race, graduation year can proxy age). 2) Measure disparate impact: selection rate ratio for protected groups should be >80% of majority group (4/5 rule). 3) Regular bias audits against diverse applicant pool. 4) Human review requirement for all AI-assisted decisions. Explainability: provide candidates with specific reasons for screening decisions. "Shortlisted for: 5+ years Python experience, data engineering background. Not shortlisted for: Java required skill not found in resume." Tools: HireEZ, Pymetrics (bias-focused), custom solution with fairlearn library for bias measurement.