Terms & Conditions

The Service provides automated screening and ranking suggestions for resumes and candidate profiles using natural language processing (NLP). It produces similarity scores, relevance rankings, and shortlists to help human reviewers. The Service relies on Sentence-BERT (SBERT) embeddings and additional algorithms to encode resume and job description text.

Data collected: resumes, CV text, cover letters, job descriptions, metadata (timestamps, uploader identity, annotations), and usage logs. Where provided, candidate contact details and attachments may be stored to support recruitment workflows.

Purpose: to compute SBERT embeddings, perform semantic matching, rank candidates, and generate shortlists for human review. We may also use anonymized/aggregated data to improve the Service and for research and analytics.

Retention: We retain personal data only as long as necessary for the purposes described or to comply with legal obligations. You may request deletion; see Section 7.

Security: We apply industry-standard technical and organizational measures (encryption in transit and at rest, access controls, logging) to protect data. However, no system is perfectly secure — see Section 6 for limits on liability.

The Service encodes text (resumes, job descriptions) into dense vector representations using SBERT (Sentence-BERT) or compatible embedding models. Similarity and ranking are computed over those vectors and combined with configurable business rules (keywords, thresholds, custom weights).

Automated scores are probabilistic signals — they are not definitive measures of candidate quality. Absolute accuracy depends on the quality of input data, model version, and configuration. We recommend human review of all shortlists before interview or hiring decisions.

Versioning: We may update or replace the embedding model (SBERT variant) over time. Model version and training data (where applicable) may affect results. Changes will be documented in our release notes and through the admin dashboard.

Automated NLP models can reflect or amplify biases present in training data. We strive to detect and mitigate biased behaviors (through preprocessing, balanced datasets, calibration, and manual audits). Nevertheless, the Service cannot guarantee bias-free outputs.

Users must ensure compliance with applicable non-discrimination laws and internal policies when using the Service. The Service provides tools and settings (e.g., anonymization, attribute filtering) to reduce bias risk, but final responsibility for fair hiring decisions rests with the human users and the organization using the Service.

THE SERVICE IS PROVIDED "AS IS" AND FOR ASSISTIVE PURPOSES ONLY. WE DO NOT GUARANTEE THAT THE SERVICE WILL IDENTIFY EVERY QUALIFIED CANDIDATE OR THAT RESULTS ARE FREE FROM ERROR OR BIAS.

To the fullest extent permitted by law, our company and its affiliates shall not be liable for indirect, incidental, special, or consequential damages arising out of your use of the Service, including hiring decisions, lost opportunities, or claims by candidates. Some jurisdictions do not allow the exclusion of certain warranties or limitations of liability — where prohibited, these limitations will not apply.

Access & correction: Candidates have the right to request access to personal data that we hold and to request corrections if data is inaccurate.

Deletion/Right to be forgotten: Candidates or account admins may request deletion of personal data. We will respond to legitimate requests within a reasonable time and remove data unless retention is required by law.

Automated decision-making: If a final decision is made solely by automated means that produces legal or similarly significant effects, eligible individuals may have rights to obtain human intervention, express their point of view, and contest the decision. We design the Service to support human-in-the-loop review to avoid solely automated adverse decisions.

You are responsible for ensuring that your use of the Service, including storage and transfer of candidate data, complies with local data protection laws (e.g., GDPR, CCPA) and employment regulations. Export controls and ML model use restrictions may apply depending on your jurisdiction.

We use encryption, network segmentation, and role-based access controls. Customers should use strong passwords, enable multi-factor authentication for admin accounts, and follow secure data handling procedures (least privilege, regular audits).

We may update these Terms from time to time. Material changes will be communicated via email and posted on our website. Continued use after notice constitutes acceptance of the updated Terms.

To request data access, deletion, or to report concerns about bias or model behavior, contact: privacy@example.com. For general questions: support@example.com.