Lý thuyết
50 phút
Bài 13/15

Recruitment AI

Tự động hóa tuyển dụng với AI - resume screening, candidate matching, và interview scheduling

Recruitment AI

Tuyển dụng truyền thống tốn 23 giờ để hire 1 người. AI có thể giảm xuống còn 8 giờ - screen resumes, match candidates, và schedule interviews tự động.

🎯 Mục tiêu bài học

  • Tự động hóa resume screening
  • Setup AI candidate matching
  • Build interview scheduling workflow
  • Ensure fair & unbiased hiring

📊 The Recruitment Challenge

Traditional Hiring Process

Text
1┌─────────────────────────────────────────────┐
2│ TRADITIONAL RECRUITMENT │
3├─────────────────────────────────────────────┤
4│ │
5│ 📋 Job Posting (2 hrs) │
6│ ↓ │
7│ 📧 Receive Applications (ongoing) │
8│ ↓ │
9│ 👀 Manual Resume Review (8 hrs) │
10│ ↓ (100 resumes × 5 min each) │
11│ ✓ Shortlist Candidates (2 hrs) │
12│ ↓ │
13│ 📞 Phone Screening (5 hrs) │
14│ ↓ (20 calls × 15 min) │
15│ 📅 Schedule Interviews (3 hrs) │
16│ ↓ │
17│ 🎯 Conduct Interviews (varies) │
18│ ↓ │
19│ ✅ Make Decision (2 hrs) │
20│ │
21│ Total: 22-25 hours per hire │
22│ Time to hire: 30-45 days │
23└─────────────────────────────────────────────┘

AI-Powered Recruitment

Text
1┌─────────────────────────────────────────────┐
2│ AI-POWERED RECRUITMENT │
3├─────────────────────────────────────────────┤
4│ │
5│ 📋 Job Posting (AI-assisted) (30 min) │
6│ ↓ │
7│ 🤖 Auto Resume Screening (instant) │
8│ ↓ (100 resumes in 2 min) │
9│ 🎯 AI Candidate Ranking (instant) │
10│ ↓ │
11│ 📅 Auto Interview Scheduling (30 min) │
12│ ↓ │
13│ 📝 AI Interview Prep (15 min) │
14│ ↓ │
15│ 🎯 Conduct Interviews (varies) │
16│ ↓ │
17│ ✅ AI-Assisted Decision (1 hr) │
18│ │
19│ Total: 6-8 hours per hire │
20│ Time to hire: 15-20 days │
21└─────────────────────────────────────────────┘

Impact: Companies using AI recruitment see 75% reduction in time-to-hire và 35% improvement in quality of hire.

🛠️ AI Recruitment Tools

Tool Comparison

ToolBest ForPriceKey Feature
ManatalSMB/Agencies$15/user/moAI recommendations
HireVueEnterpriseCustomVideo AI analysis
LeverMid-marketCustomFull ATS + AI
Paradox (Olivia)High-volumeCustomConversational AI
SeekOutTech hiring$500/moTalent sourcing
TextioJob posts$500/moInclusive writing

No-Code Options

Text
1Budget-friendly AI recruitment:
2
31. ChatGPT/Claude + Google Sheets
4 - Resume parsing
5 - Candidate scoring
6 - Email templates
7 Cost: ~$20/month
8
92. Make.com + AI
10 - Automated workflows
11 - Integration with job boards
12 - Candidate pipeline
13 Cost: ~$50/month
14
153. Manatal
16 - Full ATS with AI
17 - Resume parsing
18 - Candidate matching
19 Cost: $15/user/month

📝 AI-Powered Job Descriptions

Generate Inclusive Job Posts

Prompt:

markdown
1Write a job description for a Senior Software Engineer
2position at a fintech startup.
3
4Requirements:
5- 5+ years experience
6- Python and AWS expertise
7- Team leadership
8
9Make it:
10- Inclusive and welcoming
11- Clear about must-haves vs nice-to-haves
12- Focused on what they'll accomplish, not just requirements
13- Free of biased language

AI Output:

markdown
1# Senior Software Engineer
2
3## About the Role
4Join our engineering team to build the next generation
5of financial tools that help millions manage their money
6better. You'll lead key initiatives, mentor teammates,
7and directly impact our product roadmap.
8
9## What You'll Do
10- Design and build scalable backend systems in Python
11- Lead technical projects from concept to production
12- Mentor junior engineers and conduct code reviews
13- Collaborate with product and design teams
14- Drive best practices in code quality and testing
15
16## What We're Looking For
17
18**Must Have:**
19- 5+ years of software development experience
20- Strong Python programming skills
21- Experience with AWS services
22- Track record of shipping products
23- Clear communication skills
24
25**Nice to Have:**
26- Fintech or financial services background
27- Experience with event-driven architectures
28- Contributions to open source projects
29- Experience leading small teams
30
31## Why Join Us
32- Competitive salary + equity
33- Remote-first culture
34- Learning budget ($2,000/year)
35- Health, dental, vision insurance
36- Flexible PTO
37
38## Our Commitment to Diversity
39We believe diverse teams build better products. We
40welcome candidates from all backgrounds and are
41committed to inclusive hiring practices.
42
43---
44Location: Remote (US)
45Type: Full-time

Bias Check with AI

Text
1Prompt: "Review this job description for biased language"
2
3AI Analysis:
4┌─────────────────────────────────────────────┐
5│ JOB DESCRIPTION BIAS CHECK │
6├─────────────────────────────────────────────┤
7│ │
8│ ⚠️ Potentially Biased Language: │
9│ │
10│ "Rockstar developer" │
11│ → Replace: "Skilled developer" │
12│ → Reason: Gendered connotation │
13│ │
14│ "Young, dynamic team" │
15│ → Replace: "Collaborative team" │
16│ → Reason: Age discrimination │
17│ │
18│ "Must be native English speaker" │
19│ → Replace: "Excellent English communication"│
20│ → Reason: Nationality discrimination │
21│ │
22│ ✅ Good Practices Found: │
23│ • Clear must-have vs nice-to-have │
24│ • Focus on skills, not credentials │
25│ • Inclusive benefits mentioned │
26│ │
27└─────────────────────────────────────────────┘

📄 Resume Screening

AI Resume Parser Setup

Using Make.com + Claude:

Text
1Workflow:
21. Resume received (email/form)
32. Extract text from PDF
43. Send to Claude API
54. Parse structured data
65. Score against job requirements
76. Add to candidate database

Claude Parsing Prompt:

markdown
1Parse this resume and extract information in JSON format:
2
3{
4 "name": "",
5 "email": "",
6 "phone": "",
7 "location": "",
8 "linkedin": "",
9 "summary": "",
10 "experience": [
11 {
12 "company": "",
13 "title": "",
14 "duration": "",
15 "years": 0,
16 "responsibilities": [],
17 "achievements": []
18 }
19 ],
20 "education": [
21 {
22 "institution": "",
23 "degree": "",
24 "field": "",
25 "year": ""
26 }
27 ],
28 "skills": {
29 "technical": [],
30 "soft": [],
31 "languages": [],
32 "certifications": []
33 },
34 "total_experience_years": 0
35}
36
37Resume:
38[paste resume text]

AI Candidate Scoring

Scoring Prompt:

markdown
1Score this candidate against the job requirements.
2
3Job Requirements:
4- 5+ years Python experience (Required)
5- AWS expertise (Required)
6- Team leadership (Required)
7- Fintech experience (Preferred)
8- Open source contributions (Preferred)
9
10Candidate Profile:
11[parsed resume JSON]
12
13Provide scoring:
14{
15 "overall_score": 0-100,
16 "requirement_scores": {
17 "python_experience": {
18 "score": 0-100,
19 "evidence": "",
20 "years": 0
21 },
22 "aws_expertise": {
23 "score": 0-100,
24 "evidence": ""
25 },
26 "team_leadership": {
27 "score": 0-100,
28 "evidence": ""
29 },
30 "fintech_experience": {
31 "score": 0-100,
32 "evidence": ""
33 },
34 "open_source": {
35 "score": 0-100,
36 "evidence": ""
37 }
38 },
39 "strengths": [],
40 "concerns": [],
41 "recommendation": "Strong Match / Good Match / Weak Match / No Match",
42 "suggested_interview_questions": []
43}

Batch Processing Pipeline

Text
1Google Sheets Setup:
2├── Sheet 1: Raw Applications
3│ └── Columns: Timestamp, Name, Email, Resume Link
4
5├── Sheet 2: Parsed Candidates
6│ └── Columns: Name, Email, Experience, Skills, Score
7
8└── Sheet 3: Shortlist
9 └── Columns: Name, Score, Status, Interview Date
10
11Make.com Automation:
121. New row in Sheet 1 (trigger)
132. Download resume from link
143. Extract text (PDF.co or similar)
154. Send to Claude for parsing
165. Send to Claude for scoring
176. Add to Sheet 2
187. IF score >= 70: Add to Sheet 3
198. Send notification to recruiter

📅 Interview Scheduling

Calendly + AI Integration

Setup:

Text
11. Create Calendly account
22. Set availability for interviews
33. Create event type: "Initial Interview"
44. Get scheduling link
5
6Automation:
7Candidate passes screening →
8Auto-send email with Calendly link →
9Candidate self-schedules →
10Confirmation sent to all parties

Auto-Scheduling Email Template:

markdown
1Subject: Interview Invitation - [Position] at [Company]
2
3Hi [candidate_name],
4
5Thank you for applying to the [position] role at
6[company]. We've reviewed your application and would
7love to learn more about you!
8
9**Next Step: Schedule Your Interview**
10
11Please select a time that works for you:
12[Schedule Interview]([calendly_link])
13
14**What to Expect:**
15- 30-minute video call via Zoom
16- Discussion about your experience
17- Q&A about the role and company
18
19**Interviewer:** [interviewer_name], [interviewer_title]
20
21If none of the times work, reply to this email and
22we'll find an alternative.
23
24Best regards,
25[recruiter_name]
26[company] Talent Team

Smart Scheduling with AI

Text
1Advanced Workflow:
2
31. Candidate selected for interview
42. AI checks:
5 - Candidate timezone (from resume location)
6 - Interviewer availability
7 - Interview room availability
8 - Buffer time between interviews
93. AI suggests 3 optimal time slots
104. Auto-send to candidate
115. Candidate confirms
126. Calendar events created for all parties
137. Zoom link generated
148. Reminder emails scheduled

🤖 AI Interview Assistant

Pre-Interview Prep

Generate Interview Questions:

markdown
1Prompt: "Generate interview questions for Senior
2Software Engineer candidate based on their resume"
3
4Candidate Background:
5- 6 years at fintech companies
6- Led team of 4 engineers
7- Built payment processing system
8- Python, AWS, PostgreSQL expertise
9
10AI Output:
11
12# Interview Guide: [Candidate Name]
13
14## Technical Questions (30 min)
15
161. **System Design**
17 "Tell me about the payment processing system you
18 built. How did you handle scalability and reliability?"
19
20 Follow-ups:
21 - How did you handle failed transactions?
22 - What was your peak transaction volume?
23
242. **Python Expertise**
25 "What's your approach to writing maintainable
26 Python code at scale?"
27
28 Follow-ups:
29 - Testing strategies?
30 - Performance optimization techniques?
31
323. **AWS Architecture**
33 "Walk me through how you'd design a highly
34 available service on AWS"
35
36## Leadership Questions (15 min)
37
384. "Tell me about leading your team of 4. What was
39 your biggest challenge and how did you handle it?"
40
415. "How do you balance coding vs team responsibilities?"
42
43## Behavioral Questions (10 min)
44
456. "Describe a time when you had to push back on a
46 product requirement. What was the outcome?"
47
487. "Tell me about a project that failed. What did
49 you learn?"
50
51## Red Flags to Watch For
52- Vague answers about technical decisions
53- Unable to explain past project contributions
54- Signs of poor collaboration skills

Interview Feedback Template

markdown
1# Interview Feedback Form
2
3**Candidate:** [Name]
4**Position:** Senior Software Engineer
5**Interviewer:** [Your name]
6**Date:** [Date]
7
8## Scoring (1-5 scale)
9
10| Area | Score | Notes |
11|------|-------|-------|
12| Technical Skills | ⬜1 ⬜2 ⬜3 ⬜4 ⬜5 | |
13| Problem Solving | ⬜1 ⬜2 ⬜3 ⬜4 ⬜5 | |
14| Communication | ⬜1 ⬜2 ⬜3 ⬜4 ⬜5 | |
15| Leadership | ⬜1 ⬜2 ⬜3 ⬜4 ⬜5 | |
16| Culture Fit | ⬜1 ⬜2 ⬜3 ⬜4 ⬜5 | |
17
18## Strengths
19-
20-
21
22## Concerns
23-
24-
25
26## Recommendation
27⬜ Strong Hire
28⬜ Hire
29⬜ No Hire
30⬜ Strong No Hire
31
32## Notes for Next Interviewer
33[What should next interviewer focus on?]

📊 Candidate Pipeline Dashboard

Airtable/Notion Setup

Text
1Database: Candidates
2
3Fields:
4├── Name (text)
5├── Email (email)
6├── Position (link to Jobs)
7├── Source (select: LinkedIn, Referral, Job Board)
8├── Status (select)
9│ ├── New Application
10│ ├── AI Screening
11│ ├── Recruiter Review
12│ ├── Phone Screen
13│ ├── Interview 1
14│ ├── Interview 2
15│ ├── Offer
16│ ├── Hired
17│ └── Rejected
18├── AI Score (number)
19├── Resume (attachment)
20├── Parsed Data (JSON)
21├── Interview Notes (long text)
22├── Final Decision (select)
23└── Offer Details (text)
24
25Views:
26├── Kanban by Status
27├── Table: All Candidates
28├── Calendar: Interviews
29└── Gallery: Shortlisted

Recruitment Metrics

Text
1Dashboard Metrics:
2
3Pipeline Overview:
4├── Total Applications: 156
5├── AI Screened: 156 (100%)
6├── Passed Screening: 42 (27%)
7├── Interviewed: 18 (12%)
8├── Offers Extended: 4 (2.5%)
9└── Accepted: 3 (1.9%)
10
11Time Metrics:
12├── Avg time to screen: 2 hours
13├── Avg time to interview: 5 days
14├── Avg time to offer: 14 days
15└── Avg time to hire: 21 days
16
17Quality Metrics:
18├── Offer acceptance rate: 75%
19├── 90-day retention: 95%
20└── Hiring manager satisfaction: 4.5/5
21
22Source Performance:
23├── LinkedIn: 45 candidates, 2 hires (4.4%)
24├── Referrals: 23 candidates, 1 hire (4.3%)
25├── Job boards: 88 candidates, 0 hires (0%)

⚖️ Ethical AI Hiring

Avoiding Bias

Text
1⚠️ AI Bias Risks:
2
31. Training Data Bias
4 - Historical hiring patterns embedded
5 - May perpetuate past discrimination
6
72. Proxy Discrimination
8 - ZIP code → socioeconomic status
9 - University name → privilege
10 - Name → gender/ethnicity
11
123. Keyword Bias
13 - "Golf" vs "Basketball"
14 - "Assertive" vs "Collaborative"

Best Practices

Text
1✅ Fair AI Hiring Practices:
2
31. Blind Resume Review
4 - Remove names, photos, addresses
5 - Focus on skills and experience
6 - AI screens on requirements only
7
82. Diverse Training Data
9 - Ensure AI trained on diverse hires
10 - Regular bias audits
11 - Multiple reviewers
12
133. Structured Interviews
14 - Same questions for all candidates
15 - Scoring rubric
16 - Multiple interviewers
17
184. Human Override
19 - AI recommends, human decides
20 - Always review edge cases
21 - Appeals process available
22
235. Regular Audits
24 - Track demographics through pipeline
25 - Identify drop-off points
26 - Adjust criteria if biased

Legal Note: In many jurisdictions, AI hiring tools must comply với anti-discrimination laws. Always consult legal counsel và conduct regular bias audits.

🔧 Complete Workflow Setup

Text
1End-to-End AI Recruitment:
2
31. POST JOB
4 └── AI generates inclusive job description
5
62. RECEIVE APPLICATIONS
7 └── Auto-acknowledge receipt
8
93. AI SCREENING (Make.com)
10 ├── Parse resume (Claude)
11 ├── Score against requirements
12 ├── Flag for human review if borderline
13 └── Update candidate database
14
154. RECRUITER REVIEW
16 ├── Review AI-flagged candidates
17 ├── Final shortlist decision
18 └── Reject with personalized email
19
205. SCHEDULE INTERVIEWS
21 ├── Auto-send Calendly links
22 ├── Candidate self-schedules
23 └── Reminders automated
24
256. INTERVIEW PREP
26 ├── AI generates questions
27 ├── Interviewer gets prep doc
28 └── Candidate gets company info
29
307. POST-INTERVIEW
31 ├── Collect feedback
32 ├── AI summarizes inputs
33 └── Recommendation generated
34
358. DECISION & OFFER
36 ├── AI compiles all feedback
37 ├── Hiring manager decides
38 └── Offer letter generated (AI)
39
409. ONBOARDING TRIGGER
41 └── Accepted → Start onboarding workflow

🎯 Bài tập thực hành

Task 1: Job Description (15 phút)

Text
11. Choose a role you're hiring for
22. Use AI to generate job description
33. Run bias check
44. Refine based on feedback

Task 2: Resume Screening (30 phút)

Text
11. Create Google Sheet for candidates
22. Set up Make.com workflow:
3 - Email trigger
4 - PDF extraction
5 - Claude parsing
6 - Score calculation
73. Test with 3 sample resumes

Task 3: Interview Scheduler (20 phút)

Text
11. Set up Calendly event type
22. Create email template
33. Build automation:
4 - Candidate passes screening
5 - Auto-send scheduling link
64. Test end-to-end

📚 Tổng kết

ConceptKey Takeaway
Job PostsAI generates inclusive, clear descriptions
Screening100 resumes in minutes, not hours
ScoringObjective criteria, evidence-based
SchedulingSelf-service, automated reminders
EthicsHuman oversight, bias audits required

Tiếp theo: Bài 14 - Employee Support - Xây dựng internal Q&A bot cho employees!