Student teams propose AI résumé screeners to tackle DHRD backlog
December 01, 2025 | Enterprise Technology Services (ETS), Office of, Executive , Hawaii
This article was created by AI summarizing key points discussed. AI makes mistakes, so for full details and context, please refer to the video of the full meeting. Please report any errors so we can fix them. Report an error »
Multiple student teams presented AI tools designed to help the Department of Human Resources (DHRD) process a reported backlog of roughly 5,500 applications.
Presenters described common features: an AI scoring engine to cross-reference résumés against minimum qualifications, a reviewer dashboard for human-in-the-loop decisions, and evidence-extraction features to make AI reasoning transparent to reviewers. Teams reported small-scale tests—typically with about 10 sample résumés—yielding reported accuracy or confidence metrics in the 75–88% range, and said their systems could reduce review time from tens of minutes to seconds for initial screening.
Teams repeatedly emphasized privacy and bias mitigation. Solutions proposed redaction or pattern-matching to remove personal identifiers before AI processing, encryption of stored resumes, and manual-review thresholds for cases where the model confidence was low. One team described compliance attention to HAR chapter 76 with personal identifiable information guarded during processing.
Why it matters: DHRD and other public employers cite long application backlogs as a barrier to filling critical positions (presenters and hosts repeatedly referenced the roughly 5,500 backlog). If an AI-assisted screener can safely prioritize qualified candidates and preserve human oversight, agencies could shorten hiring timelines for critical roles such as nurses, IT specialists and administrative staff.
What presenters said about limits: Teams acknowledged limited testing data and time constraints. Several said they had run prototypes on as few as 10 sample résumés and described plans to integrate custom models, larger datasets and further bias testing before deployment. A recurring theme: "AI is an aid, not the final decision," and the final hiring call remained with human reviewers.
Next steps: Presenters invited agency partners and sponsor companies to pilot or beta-test their tools. Judges asked about bias mitigation, data sets and batch processing; teams described plans for encryption, redaction and iterative testing but did not present audited accuracy results during the session.
Presenters described common features: an AI scoring engine to cross-reference résumés against minimum qualifications, a reviewer dashboard for human-in-the-loop decisions, and evidence-extraction features to make AI reasoning transparent to reviewers. Teams reported small-scale tests—typically with about 10 sample résumés—yielding reported accuracy or confidence metrics in the 75–88% range, and said their systems could reduce review time from tens of minutes to seconds for initial screening.
Teams repeatedly emphasized privacy and bias mitigation. Solutions proposed redaction or pattern-matching to remove personal identifiers before AI processing, encryption of stored resumes, and manual-review thresholds for cases where the model confidence was low. One team described compliance attention to HAR chapter 76 with personal identifiable information guarded during processing.
Why it matters: DHRD and other public employers cite long application backlogs as a barrier to filling critical positions (presenters and hosts repeatedly referenced the roughly 5,500 backlog). If an AI-assisted screener can safely prioritize qualified candidates and preserve human oversight, agencies could shorten hiring timelines for critical roles such as nurses, IT specialists and administrative staff.
What presenters said about limits: Teams acknowledged limited testing data and time constraints. Several said they had run prototypes on as few as 10 sample résumés and described plans to integrate custom models, larger datasets and further bias testing before deployment. A recurring theme: "AI is an aid, not the final decision," and the final hiring call remained with human reviewers.
Next steps: Presenters invited agency partners and sponsor companies to pilot or beta-test their tools. Judges asked about bias mitigation, data sets and batch processing; teams described plans for encryption, redaction and iterative testing but did not present audited accuracy results during the session.
View the Full Meeting & All Its Details
This article offers just a summary. Unlock complete video, transcripts, and insights as a Founder Member.
✓
Watch full, unedited meeting videos
✓
Search every word spoken in unlimited transcripts
✓
AI summaries & real-time alerts (all government levels)
✓
Permanent access to expanding government content
30-day money-back guarantee
