AI HR & Talent Recruitment with Chinese LLMs

# Resume screening and candidate-job matching import requests job_requirements = { "position": "Senior Frontend Engineer", "required_skills": ["React", "TypeScript", "Next.js", "CI/CD"], "experience_years": "5+", "nice_to_have": ["Micro-frontends", "Performance optimization", "Team leadership"], "company_culture": "Fast-paced startup, values ownership and continuous learning", "team_size": "8 engineers, reporting to VP Engineering" } candidate = { "name": "Li Ming", "experience_years": 6, "education": "BS Computer Science, Zhejiang University", "current_role": "Frontend Tech Lead at Alibaba (3 years)", "skills": ["React", "TypeScript", "Vue.js", "Webpack", "Jest", "Docker"], "projects": "Led frontend migration from Vue to React for 2M DAU e-commerce platform. Reduced bundle size by 40%. Mentored 3 junior developers.", "career_goals": "Looking to join early-stage startup where I can have broader impact and grow into engineering management", "salary_expectation_yuan": 45000 } response = requests.post( "https://tokenease.io/v1/chat/completions", headers={"Authorization": "Bearer YOUR_TOKENEASE_API_KEY"}, json={ "model": "deepseek-v4", "messages": [ {"role": "system", "content": "You are a talent acquisition AI. Match candidates to jobs, score fit (1-100), identify strengths and gaps, assess culture fit, and recommend interview focus areas. Format as structured JSON."}, {"role": "user", "content": f"Evaluate candidate fit: Job={json.dumps(job_requirements)} | Candidate={json.dumps(candidate)}"} ], "max_tokens": 2000 } ) print(response.json()["choices"][0]["message"]["content"])

# Generate tailored technical interview questions import requests interview_context = { "role": "Backend Engineer (Go)", "seniority": "Senior (5-8 years)", "company_stack": ["Go", "gRPC", "Kubernetes", "PostgreSQL", "Redis"], "candidate_background": "Currently at ByteDance, worked on high-throughput recommendation service. Strong in distributed systems, less experience with SQL optimization.", "interview_duration_minutes": 60, "focus_areas": ["System design", "Code quality", "Problem-solving approach"], "avoid_topics": ["LeetCode hard algorithm puzzles"] } response = requests.post( "https://tokenease.io/v1/chat/completions", headers={"Authorization": "Bearer YOUR_TOKENEASE_API_KEY"}, json={ "model": "glm-4", "messages": [ {"role": "system", "content": "You are a technical interview designer. Generate relevant, practical interview questions that assess real-world engineering skills. Include follow-up questions and expected good answers. Focus on system thinking over memorization."}, {"role": "user", "content": f"Generate interview questions: {json.dumps(interview_context, ensure_ascii=False)}"} ], "max_tokens": 2500 } ) print(response.json()["choices"][0]["message"]["content"])

# Personalized 30-day onboarding plan import requests new_hire = { "name": "Zhang Wei", "role": "Product Manager", "experience": "4 years PM at traditional enterprise software, first time at internet company", "team": "Growth team, 6 engineers, 2 designers, 1 data analyst", "product": "B2C fintech app, 5M MAU, monetization via transaction fees", "strengths": ["Strong stakeholder management", "Data-driven decision making"], "gaps": ["Limited A/B testing experience", "Unfamiliar with rapid iteration culture"], "learning_style": "Prefers hands-on projects over documentation", "start_date": "2026-09-01" } response = requests.post( "https://tokenease.io/v1/chat/completions", headers={"Authorization": "Bearer YOUR_TOKENEASE_API_KEY"}, json={ "model": "qwen3-235b", "messages": [ {"role": "system", "content": "You are an onboarding specialist AI. Create detailed, personalized onboarding plans with specific tasks, learning resources, meetings, and milestones. Balance challenge with support."}, {"role": "user", "content": f"Create onboarding plan: {json.dumps(new_hire, ensure_ascii=False)}"} ], "max_tokens": 2500 } ) print(response.json()["choices"][0]["message"]["content"])

# Performance review synthesis from multiple feedback sources import requests performance_feedback = { "employee": "Wang Fang, Senior UX Designer", "review_period": "Q2-Q3 2026", "self_assessment": "I led the redesign of our mobile app which improved NPS by 15 points. I mentored 2 junior designers. I want to grow into a design lead role.", "manager_feedback": "Strong design craft and user empathy. Excellent at stakeholder communication. Needs to improve documentation consistency. Sometimes misses deadlines on complex projects.", "peer_feedback": [ "Collaborative and open to feedback. Great at design critiques.", "Sometimes changes designs last minute without communicating impact.", "Always willing to help junior team members." ], "direct_report_feedback": ["Supportive mentor, gives clear direction, encourages experimentation"], "quantitative_metrics": {"projects_completed": 8, "on_time_delivery": "75%", "design_system_contributions": 12} } response = requests.post( "https://tokenease.io/v1/chat/completions", headers={"Authorization": "Bearer YOUR_TOKENEASE_API_KEY"}, json={ "model": "deepseek-v4", "messages": [ {"role": "system", "content": "You are a performance review AI. Synthesize multi-source feedback into balanced, constructive reviews. Identify strengths, growth areas, and development recommendations. Use supportive but honest tone."}, {"role": "user", "content": f"Synthesize performance review: {json.dumps(performance_feedback, ensure_ascii=False)}"} ], "max_tokens": 2000 } ) print(response.json()["choices"][0]["message"]["content"])

# Employee pulse survey analysis and action recommendations import requests engagement_data = { "department": "Engineering (45 people)", "survey_period": "August 2026", "response_rate": "82%", "key_themes": [ "Workload has increased significantly since Q2 reorganization", "On-call rotation feels unfair - same 5 people always on-call", "Career growth paths are unclear after senior engineer level", "Appreciation for new remote-work flexibility policy", "Concerns about technical debt accumulation" ], "sentiment_scores": {"overall": 3.4, "work_life_balance": 2.8, "management": 3.6, "career_growth": 2.9, "compensation": 3.8}, "historical_trend": "Overall score declined from 4.1 (Feb) to 3.4 (Aug)", "exit_interview_themes": ["3 of last 5 exits cited burnout", "2 cited lack of growth opportunities"] } response = requests.post( "https://tokenease.io/v1/chat/completions", headers={"Authorization": "Bearer YOUR_TOKENEASE_API_KEY"}, json={ "model": "glm-4", "messages": [ {"role": "system", "content": "You are an employee engagement AI. Analyze survey data, identify root causes, predict retention risk, and recommend specific interventions with timelines. Prioritize by impact and feasibility."}, {"role": "user", "content": f"Analyze engagement data: {json.dumps(engagement_data, ensure_ascii=False)}"} ], "max_tokens": 2000 } ) print(response.json()["choices"][0]["message"]["content"])

# HR policy Q&A assistant import requests hr_query = { "employee": {"id": "EMP-4482", "department": "Sales", "tenure_months": 14, "location": "Shanghai"}, "question": "My wife is pregnant and due in December. How much paternity leave do I get, and what's the process to apply? Also, can I use my annual leave before the paternity leave starts?", "relevant_policies": [ "Paternity leave: 15 calendar days for Shanghai-registered employees, paid at full salary", "Annual leave: 10 days for employees with 1-10 years service", "Leave application: Submit via HR portal 2 weeks in advance, manager approval required", "Leave stacking: Annual leave and statutory leave can be taken consecutively with manager approval" ], "previous_similar_questions": "12 employees asked similar questions in last 3 months" } response = requests.post( "https://tokenease.io/v1/chat/completions", headers={"Authorization": "Bearer YOUR_TOKENEASE_API_KEY"}, json={ "model": "qwen3-235b", "messages": [ {"role": "system", "content": "You are an HR policy assistant AI. Answer employee questions accurately based on company policies. Be warm and supportive. Provide step-by-step guidance. Escalate complex cases."}, {"role": "user", "content": f"Answer HR query: {json.dumps(hr_query, ensure_ascii=False)}"} ], "max_tokens": 1500 } ) print(response.json()["choices"][0]["message"]["content"])