Responsible AI at DeeplyTalented

Introduction

At DeeplyTalented Inc. (“DeeplyTalented” or “us” or “we”), we believe AI should be used to empower - not replace - human judgment. As a recruiting assistant platform, we are committed to building and deploying AI responsibly, with safeguards in place to minimize bias, protect data, and ensure transparency.

This page outlines how our AI works, how we monitor its performance, and what steps we take to avoid unfair or discriminatory outcomes.

1. How We Use AI

DeeplyTalented’s AI helps recruiting teams screen and summarize candidate resumes more efficiently. It applies structured job criteria - defined by the user - to identify candidates who meet baseline qualifications (e.g., minimum years of experience, education requirements, certifications).

The AI assists with:

  • Resume parsing and structured analysis

  • Criteria-based screening support

  • Natural language summarization of candidate profiles

Importantly, our AI does not:

  • Make hiring decisions

  • Rank candidates independently

  • Use personal data for training purposes

All final decisions remain in the hands of human recruiters.

2. Preventing Bias and Discrimination

We implement multiple safeguards to reduce the risk of bias or unfair outcomes in candidate screening:

  • Human-Defined Criteria Only: Hiring teams configure all screening criteria. Our AI simply applies those rules - such as “3+ years of sales experience” or “CPA designation” - across resumes.

  • No Use of Sensitive Attributes: Our models do not infer or utilize protected characteristics such as race, gender, age, or nationality.

  • Transparent Scoring: All scorecards are created and editable by users, with full visibility into what criteria are applied. This allows users to audit and adjust as needed.

  • Fairness Testing: We periodically test our models using synthetic data to detect and address any potential disparities in how criteria are applied across varying profiles.

3. Monitoring and Oversight

We continuously monitor the performance of our AI systems to ensure reliable and consistent outputs.

  • Output Monitoring: Our engineering and product teams review model behavior to detect anomalies or inconsistent performance.

  • Human Override: Recruiters can override or adjust AI assessments at any time, ensuring the final decision always rests with a human.

  • Feedback Signals: User feedback is logged and analyzed to improve our systems - though no user data is ever used for model training.

4. Data Usage and Privacy

We take a privacy-first approach to all data handling:

  • No Training on Customer Data: We do not use client resumes or candidate data to train our AI models.

  • Public and External Sources Only: Our AI reviews publicly available candidate information (e.g., LinkedIn) or resumes submitted directly by applicants via your ATS.

  • Data Minimization: We only process what is necessary to provide our service and delete all customer data upon request or termination.

  • Industry-Standard Security: All systems operate over TLS-encrypted connections, use secure infrastructure, and undergo automated security scanning.

For full details, please refer to our Platform Privacy & Security Policy or our Software and Data Processing Agreement (available upon request).

5. Security Practices

We employ technical and organizational measures aligned with industry best practices, including:

  • TLS encryption for all in-transit data

  • Firewall-protected cloud infrastructure

  • Key-based SSH access control

  • Role-based access permissions

  • Continuous vulnerability scanning as part of CI/CD

  • Cloud hosting on SOC 2 Type II-certified infrastructure

6. Subprocessor Transparency

We partner with a small number of trusted subprocessors to support our AI and infrastructure. Each is contractually bound to meet our security and privacy standards.

We do not sell, share, or retain personal data beyond what is required to deliver our services.

7. Contact

We welcome any questions or feedback about our approach to AI governance and fairness.

Please contact us at [email protected] or through your customer success representative.