AI Recruitment Masterclass

Take one open role from intake meeting to signed offer, using AI at every step.

12modules
73lessons
10labs
~4hreading
500resume dataset

Most AI recruiting courses hand you a list of a thousand tools. This one hands you a pipeline.

You take a single open role from intake meeting to signed offer using AI at every step — writing the job profile, sourcing a longlist, screening several hundred resumes, ranking a shortlist you can defend, running structured interviews, automating the handoffs, and reporting the result. Every lab runs on the same 500-resume dataset, so nothing is theoretical.

By the end you will have a job blueprint, a scorecard, a sourcing prompt library, a screened shortlist, an interview loop, a configured pipeline, a weekly dashboard and an AI hiring risk register. Those are artefacts, not notes.

Start Module 1, Lesson 1 See a real run first

No lessons marked read yet
— stored in this browser only

Short on time? Three paths through it

PathLengthFor
Founder12 lessons, ~2hHiring without a recruiting function
Agency & staffing13 lessons, ~3hHigh requisition volume across clients
Compliance & HR leadership14 lessons, ~2.5hAccountable for AI in hiring without operating it

Who this is for

  • Recruiters and talent acquisition specialists who have been told to "use AI" and want a method rather than a tool list.
  • Agency and staffing recruiters carrying high requisition volume.
  • Founders and hiring managers hiring without a recruiting function.
  • Anyone responsible for AI hiring compliance who needs to know what the tools actually do.

What it does not cover

Executive search, campus hiring at scale, or building your own models. It teaches one repeatable pipeline for professional hiring, in depth, with tools that exist today. The emphasis is on the decisions — what to weight, what to check, what never to automate — rather than on any single interface.

Curriculum

12 modules · 73 lessons · 10 labs. Open any lesson to see what it covers.

1

Module 1 · Foundations of AI recruiting

What AI actually changes, the seven-layer stack, the five failure modes, and your workspace setup.

5 lessons  ·  21 min
  • 1What AI actually changes in the hiring funnel4 min
    • How the funnel keeps its shape while the labour moves upstream
    • Which stages AI compresses, and the two that stay entirely human
    • The volume asymmetry: applications rising, recruiting headcount flat
    • What you are actually buying: coverage, consistency, an audit record
    • Why vague criteria produce a fast, consistent, well-documented bad shortlist
    Read this lesson
  • 2The AI recruiting stack: seven layers4 min
    • The seven layers: record, ingestion, search, matching, communication, orchestration, governance
    • Why layer 2 is load-bearing and never appears in a sales demo
    • Spotting a single-layer tool being sold as a stack
    • Where the market is consolidating and where to stay flexible
    • Using the layer map as a shopping list on vendor calls
    Read this lesson
  • 3Where AI fails: invention, proxy bias, over-filtering4 min
    • Confident invention: how models fill gaps in extraction and summaries
    • Proxy bias: postcode, institution, gap length and phrasing as demographic correlates
    • Over-filtering, and why false negatives generate no feedback at all
    • Drift in criteria, in the market, and in vendor models
    • Automation complacency, and checkpoints that resist it
    Read this lesson
  • 4Build, buy, or bolt onto your ATS4 min
    • Cost, control and time profiles of the three routes
    • When building your own is genuinely defensible
    • Four questions that end a vendor demo call quickly
    • Why bolt-on quality is a function of your existing data hygiene
    • A four-question decision checklist
    Read this lesson
  • 5Set up your workspace and course dataset5 min
    • What is in the 500-resume dataset, and why it is deliberately messy
    • Creating a screening workspace on a free plan
    • Bulk upload and confirming the parse completed
    • The parsed-fields view, and why a score-only tool will not work here
    • First exercise: read three parsed outputs against their originals
    Read this lesson
2

Module 2 · Intake and the job blueprint

Turning an intake conversation into a weighted profile, a scorecard, and criteria you can screen against.

6 lessons  ·  22 min
  • 1The intake meeting: 12 questions that prevent bad hires4 min
    • The twelve questions, and the three that matter most
    • Getting a hiring manager to describe outcomes rather than a person
    • Surfacing the unstated must-have before it appears at offer stage
    • Pinning down the trade-off: what would you give up to get this
    • Recording intake in a form the rest of the pipeline can use
    Read this lesson
  • 2Turning messy intake notes into a job profile with AI3 min
    • Turning a recorded intake into structured requirements
    • Prompts that extract what was said without inventing what was not
    • Separating stated requirements from your own inferences
    • Getting sign-off on the profile before sourcing starts
    Read this lesson
  • 3Job descriptions that convert and rank4 min
    • What actually drives application rate, and what does not
    • Writing for search: titles, skills and location phrasing
    • Removing requirement inflation that suppresses applications
    • Phrasing that widens the pool without lowering the bar
    • Three templates: technical, commercial, operational
    Read this lesson
  • 4Scorecards and competency models4 min
    • What a scorecard is, and why it is not a list of skills
    • Competency definitions written as observable behaviours
    • Rating anchors, so two interviewers mean the same thing by 3
    • Tying every scorecard line to a stage that will actually assess it
    Read this lesson
  • 5Must-have vs nice-to-have: the weighting exercise4 min
    • The three-must-have discipline
    • Converting nice-to-haves into weighted signals
    • Assigning weights you can defend to a hiring manager
    • Testing your weights against two real past hires
    Read this lesson
  • 6Lab 1: build a full job blueprint3 min
    • Build a job blueprint from the dataset brief
    • Produce a scorecard with anchors and weights
    • Deliverable and success criteria
    • Self-check questions before you start sourcing
    Read this lesson
3

Module 3 · Sourcing with AI

Boolean, X-Ray, LLM query expansion, talent mapping and a reusable prompt library.

8 lessons  ·  27 min
  • 1Sourcing math: profiles per hire, by role type4 min
    • Working backwards from hires to profiles needed
    • Response and conversion benchmarks by role type
    • Why the ratio changes for niche and senior roles
    • Using the math to reset a hiring manager's expectations
    Read this lesson
  • 2Boolean that still works3 min
    • Operators that still work across the major platforms
    • Query structure: core terms, synonyms, exclusions
    • Boolean mistakes that silently narrow your results
    • Writing queries you can version and reuse
    Read this lesson
  • 3X-Ray search: LinkedIn, GitHub, Behance, Stack Overflow3 min
    • Site-restricted search on LinkedIn, GitHub, Behance and Stack Overflow
    • Platform-specific structures worth targeting
    • Finding candidates who are invisible on LinkedIn
    • Where X-Ray results go stale, and how to tell
    Read this lesson
  • 4Using LLMs to expand titles, skills and synonyms3 min
    • Generating title variants across companies and countries
    • Expanding a skill into tool, framework and synonym sets
    • Prompts that return search strings rather than prose
    • Validating an expansion against real results before trusting it
    Read this lesson
  • 5Talent mapping and market intelligence3 min
    • Mapping where your target talent actually sits
    • Estimating pool size and competitive density
    • Compensation and location signals from public data
    • Turning a map into a sourcing plan and a difficult conversation
    Read this lesson
  • 6Building your reusable sourcing prompt library3 min
    • Anatomy of a reusable prompt: role, input, constraints, output format
    • Twenty prompts across profiling, expansion, outreach and summarising
    • Versioning prompts so you can tell what changed and when
    • Testing a prompt before it enters the library
    Read this lesson
  • 7Scraping, parsing and the legal line4 min
    • What scraping tools actually do, and where they break
    • Terms of service, rate limits and account risk
    • Data-protection duties once you store a sourced profile
    • Practical rules that keep you on the right side of the line
    Read this lesson
  • 8Lab 2: a 50-profile longlist in 40 minutes4 min
    • Build a 50-profile longlist for the dataset role
    • A time-boxed workflow with checkpoints
    • Deliverable: the longlist plus the queries that produced it
    • Self-check: could someone else reproduce your search
    Read this lesson
4

Module 4 · Resume parsing and AI screening

The anchor module: how parsing works, why fields matter more than scores, and a full 500-resume run.

7 lessons  ·  31 min
  • 1How resume parsing actually works6 min
    • The seven-stage pipeline from file intake to field mapping
    • Where OCR, layout reconstruction and segmentation break
    • Why freelance and concurrent roles produce duplicate or missing positions
    • Why a single “99% accurate” figure tells you nothing
    • Worst-case exercise: the scanned resumes in the dataset
    Read this lesson
  • 2Why field accuracy decides everything downstream5 min
    • How one bad field becomes an invisible rejection
    • Fields ranked by blast radius, and why dates matter most
    • Running a 25-resume field audit in thirty minutes
    • Reading the results: thresholds for filtering versus weighting
    • When a layout failure becomes a fairness problem
    Read this lesson
  • 3Setting screening criteria that do not over-filter5 min
    • Hard filters versus weighted signals
    • Six things you should never hard-filter on, and what to use instead
    • Converting a scorecard into must-haves, signals and interview topics
    • The three-must-have discipline
    • Calibrating on 50 before you run on 500
    Read this lesson
  • 4Screening 500 resumes: the full walkthrough4 min
    • Confirming ingestion and catching files that never parsed
    • Spot-checking the parse before you enter any criteria
    • Reading the score distribution: gradients, cliffs and flat pools
    • The disagreement check: three highest rejects, three lowest accepts
    • Drawing the line by interview capacity, and recording the criteria version
    Read this lesson
  • 5Reading AI screening scores critically4 min
    • What a screening score is, and what it is not
    • Precision theatre, and why banding beats ranking
    • Three tests for whether an explanation is real
    • Score distributions that indicate a problem
    • Format correlations masquerading as hiring signal
    Read this lesson
  • 6Edge cases: gaps, career changers, non-linear CVs3 min
    • Career gaps, and why they should not be screened on
    • Career changers and the full-text second pass
    • Contractors, concurrent roles and tenure-sensitive scoring
    • Internal, referred and international candidates
    • Very senior candidates and truncated histories
    Read this lesson
  • 7Lab 3: screen the dataset, compare to a human shortlist4 min
    • Screen all 500 and produce a shortlist of twenty
    • Compare against the human shortlist supplied with the dataset
    • Analysing four groups: overlap, yours only, theirs only, neither
    • What healthy agreement looks like, and what perfect agreement means
    • Deliverable and self-check
    Read this lesson
5

Module 5 · Candidate matching and ranking

Semantic matching, match profiles, silver medallists and defensible explanations.

5 lessons  ·  19 min
  • 1Keyword match vs semantic match vs learned ranking4 min
    • What keyword matching can and cannot see
    • How semantic matching represents meaning, and where it drifts
    • Learned ranking and the historical bias it inherits
    • Choosing the method that suits your role and your pool
    Read this lesson
  • 2Building a match profile that reflects the scorecard4 min
    • Translating a scorecard into a match profile
    • Weighting evidence types: employment, projects, education, free text
    • Avoiding a profile that simply describes your last hire
    • Testing a profile against known-good and known-bad candidates
    Read this lesson
  • 3Silver medallists: re-matching your existing database4 min
    • Why previous applicants are your cheapest source
    • Re-matching an existing database against a new role
    • Keeping re-engagement compliant and welcome
    • Measuring silver-medallist conversion separately
    Read this lesson
  • 4Explainability: why did this candidate rank third?3 min
    • What a defensible explanation contains
    • Testing an explanation by removing the evidence it cites
    • Presenting rank order without over-claiming precision
    • Banding instead of ranking, and the questions it stops
    Read this lesson
  • 5Lab 4: rank and justify a shortlist of ten4 min
    • Rank ten candidates and write a justification for each
    • Compare your ordering with the system's
    • Deliverable: a ranked list with cited evidence
    • Self-check: would this survive a challenge
    Read this lesson
6

Module 6 · Outreach and candidate communication

Personalisation at scale, five-touch sequences, chatbots and rejections that protect your brand.

6 lessons  ·  20 min
  • 1Why most recruiter outreach is ignored4 min
    • Why volume outreach stopped working
    • What candidates actually screen a message on in four seconds
    • Reply-rate benchmarks, and measuring your own honestly
    • The two failures: irrelevance, and obvious templating
    Read this lesson
  • 2Personalisation at scale that does not read like a bot3 min
    • Personalisation built on evidence rather than flattery
    • Assembling message inputs from profile data
    • Prompts that vary substance, not adjectives
    • QA rules that catch the tells before you send
    Read this lesson
  • 3The five-touch sequence4 min
    • Structure and timing of a five-touch sequence
    • What each touch is for, and what changes between them
    • When to stop, and how to stop well
    • Measuring by step rather than by campaign
    Read this lesson
  • 4Chatbots for candidate FAQs and pre-qualification3 min
    • What a candidate chatbot should and should not answer
    • Pre-qualification that does not become hidden screening
    • Escalation to a human, and how fast it has to be
    • Disclosure duties when a bot handles candidates
    Read this lesson
  • 5Rejections that protect your employer brand3 min
    • Why silence costs more than rejection
    • Rejection templates by stage
    • Giving feedback without creating legal exposure
    • Keeping rejected candidates open to future roles
    Read this lesson
  • 6Lab 5: write and QA a full sequence3 min
    • Write a full five-touch sequence for the dataset role
    • Run the template-tell QA checklist against it
    • Deliverable: the sequence plus a measurement plan
    • Self-check: would you reply to touch one
    Read this lesson
7

Module 7 · Structured and AI interviews

Question banks, async interviews, scheduling, transcripts to evidence, and debrief facilitation.

7 lessons  ·  22 min
  • 1Structured interviewing: what the evidence says4 min
    • Why structure improves predictive validity
    • What an unstructured interview actually measures
    • The four components: same questions, anchors, independent scoring, structured debrief
    • Hiring-manager objections, and answers that work
    Read this lesson
  • 2Generating competency question banks3 min
    • Generating questions from competencies rather than job titles
    • Behavioural, situational and work-sample question types
    • Writing probes and follow-ups in advance
    • Removing questions that reward rehearsal
    Read this lesson
  • 3Async and AI interviews: when they help, when they hurt4 min
    • Where async video helps: scheduling, scale, time zones
    • Where it hurts: drop-off, accessibility, candidate experience
    • AI scoring of video, the highest-risk feature in the stack
    • Regulatory exposure specific to automated assessment
    Read this lesson
  • 4Interview scheduling automation3 min
    • Removing scheduling latency from your funnel
    • Panel and multi-stage coordination
    • Automated rescheduling without losing the candidate
    • What still needs a human
    Read this lesson
  • 5From transcript to evidence to scorecard3 min
    • Recording and transcription consent
    • Extracting evidence rather than summaries
    • Mapping evidence onto scorecard lines
    • Guarding against transcript hallucination
    Read this lesson
  • 6Debrief facilitation and bias interrupts2 min
    • Running a debrief that is not an anchoring exercise
    • Independent scoring before any discussion
    • Bias interrupts that work in a real meeting
    • Documenting the decision
    Read this lesson
  • 7Lab 6: run one structured interview loop3 min
    • Run one structured loop end to end
    • Produce scorecards with cited evidence
    • Deliverable: loop design plus completed scorecards
    • Self-check: could a second interviewer replicate your rating
    Read this lesson
8

Module 8 · ATS and recruitment CRM operations

Pipeline design, migration, nurture, career sites and job distribution.

6 lessons  ·  22 min
  • 1ATS anatomy: stages, statuses, automations, permissions4 min
    • Stages, statuses, and what each is actually for
    • Permissions, and why they matter for compliance
    • Automations worth enabling on day one
    • Designing a pipeline you will keep clean
    Read this lesson
  • 2Migrating off spreadsheets without losing history4 min
    • Auditing what your spreadsheet actually contains
    • Field mapping and deduplication before import
    • Preserving history and source attribution
    • Verifying the migration before switching the old process off
    Read this lesson
  • 3Recruitment CRM: nurturing before the req exists4 min
    • The difference between an ATS and a recruitment CRM
    • Building talent pools before requisitions open
    • Nurture cadence that does not become spam
    • Measuring pipeline built, not emails sent
    Read this lesson
  • 4Career site and job board setup4 min
    • Career site structure, and what candidates look for first
    • Job page markup and search visibility
    • Application form length and drop-off
    • Accessibility requirements you cannot skip
    Read this lesson
  • 5Job distribution and multi-posting3 min
    • Where to post, and how to decide
    • Multi-posting mechanics and duplicate handling
    • Tracking source properly so channel ROI is measurable
    • Paid versus organic distribution
    Read this lesson
  • 6Lab 7: configure a pipeline end to end3 min
    • Configure stages, automations and a career page
    • Run one test candidate through the entire pipeline
    • Deliverable: configuration plus test log
    • Self-check: where would a real candidate get stuck
    Read this lesson
9

Module 9 · Agents and automation

From prompts to workflows to agents, with human-in-the-loop checkpoints that actually catch errors.

6 lessons  ·  21 min
  • 1The ladder: prompt, workflow, agent4 min
    • Prompt, workflow and agent: what actually differs
    • Choosing the lowest rung that solves the problem
    • Where agents genuinely earn their complexity
    • Why the failure surface grows with autonomy
    Read this lesson
  • 2Writing an AI recruiting agent brief4 min
    • Writing an agent brief: goal, tools, limits, stop conditions
    • Defining what the agent may never do
    • Test cases before deployment
    • Ownership and review cadence
    Read this lesson
  • 3Human-in-the-loop checkpoints that catch errors4 min
    • Checkpoints that require a decision, not a click
    • Where to place review in a screening-to-scheduling flow
    • Detecting silent failures
    • Logging that makes an incident reconstructable
    Read this lesson
  • 4Connecting ATS, calendar, email and Slack3 min
    • Integration patterns across ATS, calendar, email and chat
    • Authentication and data-scope decisions
    • Handling partial failure across systems
    • Keeping the system of record authoritative
    Read this lesson
  • 5Cost control and failure modes3 min
    • Per-candidate cost modelling
    • Rate limits, retries and runaway loops
    • What breaks first at volume
    • Kill switches and rollback
    Read this lesson
  • 6Lab 8: build screening-to-scheduling automation3 min
    • Build screening-to-scheduling automation with a human checkpoint
    • Test it with deliberately broken inputs
    • Deliverable: the workflow plus a failure log
    • Self-check: what happens when the calendar is full
    Read this lesson
10

Module 10 · Recruiting analytics and KPIs

Funnel models, the four metrics that matter, source ROI and a weekly dashboard.

6 lessons  ·  24 min
  • 1The funnel model every recruiting report is built on6 min
    • Defining stages so the funnel is measurable at all
    • Conversion rates that matter, and the vanity metrics
    • Diagnosing where a funnel is actually failing
    • Reporting cadence
    Read this lesson
  • 2Time to hire, cost per hire, quality of hire4 min
    • Time to hire versus time to fill, and which to use when
    • Cost per hire: what to include, what to leave out
    • Quality of hire: proxies that are honest about being proxies
    • Presenting all three without over-claiming
    Read this lesson
  • 3Source effectiveness and channel ROI4 min
    • Attributing hires to sources properly
    • Cost and quality by channel, not just volume
    • Deciding where to stop spending
    • Common attribution errors
    Read this lesson
  • 4Forecasting hiring capacity4 min
    • Modelling recruiter capacity against a hiring plan
    • Forecasting requisition load
    • Where AI changes the capacity math, and where it does not
    • Communicating a capacity limit to the business
    Read this lesson
  • 5Building a weekly hiring dashboard3 min
    • The five numbers a weekly dashboard needs
    • Building it from ATS exports
    • Making it readable by people who are not recruiters
    • Keeping it current without manual work
    Read this lesson
  • 6Lab 9: build your dashboard3 min
    • Build a weekly dashboard from the dataset funnel
    • Deliverable: dashboard plus three sentences of commentary
    • Self-check: what decision would this actually change
    Read this lesson
11

Module 11 · Compliance, bias and candidate experience

The regulatory map, running a bias audit, disclosure language, retention and documentation.

6 lessons  ·  22 min
  • 1The regulatory map: EU AI Act, NYC LL-144, GDPR, DPDP5 min
    • EU AI Act: why recruitment sits in the high-risk category
    • NYC Local Law 144 and the bias-audit requirement
    • GDPR and India's DPDP duties for candidate data
    • Working out which regimes apply to your pipeline
    • Keeping up as the rules move
    Read this lesson
  • 2Running a bias audit in practice4 min
    • What a bias audit actually measures
    • Selection rates and impact ratios
    • Auditing parse quality as a fairness question
    • Who can run it, and how it gets documented
    Read this lesson
  • 3Disclosure and consent language3 min
    • When you must tell candidates AI is involved
    • Disclosure that is clear rather than defensive
    • Consent for recording and automated assessment
    • Three notice templates
    Read this lesson
  • 4Data retention and candidate rights3 min
    • Retention periods and lawful basis
    • Access, correction and deletion requests
    • Where candidate data leaks in an AI stack
    • Vendor data-processing terms to check
    Read this lesson
  • 5The documentation your legal team will ask for3 min
    • The documentation set legal will ask for
    • Recording criteria versions alongside decisions
    • Handling a complaint about an automated decision
    • Keeping the record without creating new risk
    Read this lesson
  • 6Lab 10: complete an AI hiring risk register4 min
    • Complete an AI hiring risk register
    • Rate likelihood and impact for each entry
    • Deliverable: register plus three mitigations you will implement
    • Self-check: which risk are you currently accepting
    Read this lesson
12

Module 12 · Rollout and capstone

Your 30-60-90 plan, hiring manager adoption, the ROI case, and the end-to-end capstone.

5 lessons  ·  17 min
  • 1Your 30-60-90 rollout plan4 min
    • What changes in the first 30, 60 and 90 days
    • Sequencing so early wins fund the later changes
    • What not to touch in month one
    • Measuring the rollout itself
    Read this lesson
  • 2Getting hiring managers to actually use it3 min
    • Why hiring managers resist, specifically
    • Making the scorecard the shared artefact
    • Training that fits on one page
    • Handling the manager who overrides everything
    Read this lesson
  • 3The business case and ROI model for your CFO3 min
    • Building the ROI model: time, cost, quality, risk
    • Numbers a CFO will accept, and ones they will not
    • Framing the ask against the real alternative
    • Presenting uncertainty honestly
    Read this lesson
  • 4Capstone: run one requisition end to end3 min
    • Capstone brief: run one real requisition end to end
    • Artefacts to produce at each stage
    • How to evaluate your own run
    • What to change before the second one
    Read this lesson
  • 5Your AI recruiting toolkit and next steps4 min
    • What you built across the course
    • The template pack and where to get it
    • Where to take this next
    • Running the same pipeline at volume
    Read this lesson

Common questions

Is this AI recruitment course free?

Yes. All 73 lessons, the 500-resume practice dataset, and the twenty-one templates are free and ungated — no account, no email required. The templates are licensed for reuse including inside your own training material.

Who is this course for?

In-house recruiters and talent acquisition specialists, agency and staffing recruiters, founders hiring without a recruiting function, and anyone accountable for AI hiring compliance. There are three shorter curated paths if the full course is more than you need.

Do you need technical knowledge to take it?

No. It assumes you have screened resumes or run interviews before, and nothing else. The most technical material — parsing internals and automation design — is explained from first principles because those are exactly the layers recruiters are usually asked to trust blindly.

How long does the course take?

About four hours of reading across 73 lessons, plus ten labs that take between forty minutes and ninety minutes each. Most lessons are four to six minutes. You can do the reading without the labs, but the labs are where the method becomes yours.

What makes this different from other AI recruiting courses?

It teaches one repeatable pipeline in depth rather than surveying a thousand tools, every lab runs on the same real dataset, and it includes a published worked run where the author's own criteria failed twice — with the numbers. It is also explicit about its conflict of interest: the author builds recruitment software.