Abstract

Automated hiring systems increasingly rely on opaque screening algorithms that rank or filter applicants prior to human review. While these systems offer operational efficiency, they raise significant concerns related to fairness, transparency, and consent. This paper describes a governance-constrained portfolio hiring architecture built on the QGI framework. The system replaces algorithmic screening with structured scenario-based evaluation, generating a shared portfolio of candidate reasoning artifacts while ensuring that AI-generated prompts satisfy deterministic governance constraints.

1. Background

Digital hiring systems have traditionally focused on efficiency: parsing resumes, ranking applicants, and automating early stages of candidate selection. Many such systems employ machine learning models trained on historical hiring data or behavioral signals.

However, these methods introduce several governance challenges:

• opaque evaluation criteria
• limited transparency for applicants
• potential bias inherited from training data
• limited accountability for automated filtering

As regulatory attention toward automated decision systems increases, organizations must reconsider how AI should participate in hiring processes.

The QGI Portfolio Hiring architecture explores a model where AI assists evaluation but does not perform autonomous candidate selection.

2. Architectural Principles

The system is designed around three core principles:

• Transparency of evaluation artifacts. Candidate responses should produce durable reasoning artifacts accessible to both parties.
• Governance before automation. AI-generated prompts must satisfy governance constraints prior to candidate interaction.
• Human decision authority. Final hiring decisions remain entirely human-driven.

3. Portfolio Evaluation Framework

Instead of evaluating resumes alone, the system constructs a candidate portfolio based on responses to contextual scenarios.

Employers configure evaluation prompts across eight dimensions:

1. Capability to perform the role
2. Decision-making and judgment
3. Cultural and environmental alignment
4. Learning capacity and adaptability
5. Integrity and responsibility
6. Communication and collaboration
7. Motivation and intent
8. Autonomy and self-management

These dimensions reflect the practical factors organizations typically evaluate when selecting candidates.

4. Scenario Generation

Using employer inputs, the system dynamically generates contextual scenarios and problem prompts relevant to the role. Examples may include:

• decision trade-offs in operational situations
• collaboration challenges within team environments
• ethical reasoning in ambiguous circumstances
• prioritization tasks under resource constraints

Candidates respond in written form, explaining their reasoning and decision process.

5. QGI Governance Gate

Before any prompt is delivered to applicants, it passes through the QGI governance layer. This step ensures that generated scenarios satisfy governance constraints including:

• fairness and non-discrimination
• transparency of evaluation context
• respect for applicant autonomy
• avoidance of harmful or inappropriate prompts

By enforcing governance constraints at the generation stage, the system prevents problematic evaluation logic from entering the hiring process.

Tier 1 Parameters. The following parameters are applied when the system operates in the hiring domain:

6. Portfolio Construction

Candidate responses are aggregated into a structured portfolio containing:

• scenario descriptions
• candidate reasoning and explanations
• contextual evaluation artifacts

This portfolio provides evaluators with insight into how candidates analyze problems, communicate decisions, and frame complex situations.

The finalized portfolio is shared by both the company and the candidate.

7. Human Decision Layer

The system does not generate automated hiring decisions. Instead, it presents the portfolio to human reviewers, who may consider the responses alongside traditional materials such as resumes and interviews.

The architecture therefore supports human-centered hiring while improving the quality of information available to evaluators.

8. Implications

Governance-constrained portfolio hiring demonstrates how AI can support hiring processes without replacing human judgment. By shifting the focus from automated filtering to structured reasoning evidence, organizations can create more transparent and accountable hiring systems.

This model illustrates one possible application of governance-first AI design in sensitive decision domains.