This department teaches organisations how to use AI to screen talent reliably — and to do so without falling for hype, biased shortcuts, or invisible decision-making that exposes institutions to legal, ethical, and operational risk. We take a pragmatic, evidence-first approach: AI is a tool to augment assessment, not replace human judgement. Our work is about measurable improvement in selection accuracy, time-to-hire, and fair outcomes.

Mission and outcomes

Enable HR teams, hiring managers, and talent professionals to design, deploy, and govern AI-based screening systems that actually work in production: faster shortlisting, better match-to-role, defensible selection decisions, and demonstrable bias mitigation. If an AI model can’t show improved hiring outcomes in a pilot, we don’t promote it.

Who this is for

• HR leaders and talent acquisition teams wanting to modernise screening.

• Technical teams (data scientists / ML engineers) building or integrating ATS/AI systems.

• Legal, compliance and privacy officers who must control risk.

• Line managers who make hiring decisions and need transparent, usable candidate outputs.

• Small and medium enterprises seeking scalable assessment without overpaying for black-box products.

Services and programmes

We offer a mix of consulting engagements and practical training modules:

Consulting (tailored engagements)

• Screening pipeline design — from job profile to candidate ranking: requirements, features, data sources, and score interpretation.

• Vendor & tool evaluation — objective audits of off-the-shelf screening products against performance, explainability, and compliance.

• PoC & pilot implementation — build, test, and measure pilots (end-to-end) with real hiring workflows and KPIs.

• Bias & fairness audits — statistical tests, subgroup performance checks, and remediation plans.

• Governance & policy setup — decision rules, human-in-the-loop gates, appeal procedures, audit logs, and retention policies.

• Integration & deployment — ATS connectors, API designs, deployment strategy, monitoring and retraining schedule.

Training (modular courses & workshops)

• Foundations for HR — what AI can/can’t do; reading model outputs; avoiding common pitfalls.

• Technical basics for practitioners — model types (NLP for CV parsing, classification models for fit scoring), feature engineering, evaluation metrics.

• Bias mitigation & fair design — practical techniques, proxy identification, balanced sampling, fairness-aware modeling.

• Privacy & legal compliance (GDPR-focused) — lawful bases for processing, data minimisation, DSARs, retention and purpose limitation.

• Explainability & communication — how to explain model decisions to candidates and hiring managers; creating human-friendly scorecards.

• Operational workshops — creating SLAs for screening systems, monitoring pipelines, drift detection, and response playbooks.

• Hands-on labs — sandboxed datasets, simulated ATS integration, model evaluation exercises, and A/B pilot design.

Each training module can be delivered as a public short course, in-house workshop (half-day to multi-day), or a certified executive programme with assessment and badge.

Methods and technology overview

We focus on proven, explainable approaches:

• NLP-based résumé parsing and feature extraction (keyword + contextual embeddings when appropriate).

• Supervised classification / ranking models trained on validated historical outcomes (performance, retention, promotion) — never on biased proxies like unvetted past hiring decisions alone.

• Hybrid rule + model pipelines: deterministic filters for regulatory or mandatory checks, followed by model scoring.

• Human-in-the-loop gating: automated shortlist, human review points, and clear escalation rules.

• Monitoring & feedback loops: outcome tracking (hire/no-hire, performance), drift detection, periodic recalibration.

We avoid black-box vendor promises unless accompanied by accessible model cards, documented evaluation, and independent audits.

Ethics, fairness and legal compliance

Operating in France and the EU requires strict adherence to data protection and anti-discrimination regimes. We insist on:

• GDPR compliance: lawful processing, purpose limitation, data minimisation, secure storage, and clear retention schedules.

• Transparency and candidate rights: meaningful information about automated decisions, opt-out and human-review channels, and clear appeal mechanisms.

• Bias identification and mitigation: statistical tests across protected groups, removal of proxies, and use of fairness-aware thresholds.

• Documentation and auditability: model cards, training data provenance, performance metrics by subgroup, and change logs for retraining.

• Ethical design principles: accountability owners, human override, and proportionate use of automation.

We make no claim that AI will be bias-free; instead, we provide processes to measure, reduce, and manage bias while protecting institutional integrity.

Practical training & partnerships

Hands-on experience matters. The department works with partner employers, ATS vendors, and hiring panels to run real pilots and internships:

• Live pilots with partner organisations to validate pipelines on current vacancies.

• Sandbox ATS environment for students and clients to test connectors, simulate candidate flows, and verify monitoring.

• Problem-led capstone projects where trainees deliver a deployable pilot and evaluation report.

Faculty and instructors

Instructors combine HR practice, data science, legal/compliance expertise, and operational engineering. Expect practitioners who have implemented screening systems, audited vendor models, or led recruitment operations in regulated environments. We value demonstrable outcomes: instructors are selected based on projects they’ve executed and problems they’ve actually solved.

Deliverables & measurable KPIs

For consulting engagements we provide clear deliverables:

• Requirements document and screening architecture.

• Model performance report (precision/recall, calibration, subgroup metrics).

• Bias audit report and remediation plan.

• Deployment playbook, monitoring dashboard, and retraining schedule.

• Candidate-facing disclosure templates and appeals workflow.
  KPIs to track (examples): time-to-hire, shortlist-to-hire conversion, first-year retention, candidate satisfaction, selection error rates by subgroup, and model drift indicators.

Risk management & limitations

We’re blunt about what can go wrong and how to mitigate it:

• Garbage-in → garbage-out: poor or unrepresentative training data produces bad models; we insist on data audits.

• Proxy risks: seemingly neutral features often proxy for protected traits — we test for and remove such proxies.

• Over-reliance on automation: automation should assist, not replace, human judgement in borderline or high-stakes roles.

• Regulatory exposure: automated adverse impact without adequate governance can lead to sanctions and reputational damage.

Our contracts include explicit success criteria for pilots and staged rollouts to limit exposure.

Certification & evaluation

We offer a certification path for practitioners who complete core modules and pass practical assessments: Certified AI Talent Screener — Practitioner. Certification requires delivering a small pilot and demonstrating responsible deployment practices.

Pricing & engagement models (summary)

• Workshops: half-day to multi-day, priced per head or flat fee for in-house sessions.

• Consulting projects: fixed-fee discovery + milestone payments for PoC, pilot, and rollout phases.

• Ongoing support: retainer models for monitoring, model maintenance, and periodic audits.
  We tailor scope to organisation size and regulatory complexity.

Why this department is different

We don’t sell shiny “set-it-and-forget-it” systems. We demand measurable improvement, full transparency, and defensible processes. The goal is a robust, auditable screening capability that reduces hiring waste, improves match quality, and protects your organisation from legal and reputational risk.