Head of Research & Development

About the role

MaacVerify, Inc. is an independent AI performance verification and certification company. We assess AI deployments inside client environments in regulated sectors (healthcare, financial services, legal) using the MAAC framework, a nine-dimension psychometric instrument validated through doctoral research in Industrial and Systems Engineering. We do not build, sell, or train AI systems, and we do not advise on governance, compliance, or remediation. We measure AI performance and issue certification decisions. Independence is the product. The company is pre-revenue and founder-led, with the methodology, research base, intellectual property, and platform in place. This is a founding team position: high ownership, high accountability, compensated in founding equity under the structure described below. The Head of Research and Development owns the continuity, extension, and scientific integrity of the MAAC research program. The instrument (MAAC v4.7) is validated through a four-paper research series; three preprints carry Zenodo DOIs and a fourth manuscript is under journal review. Your mandate is to carry that program forward: execute the pre-registered multi-model and criterion-validity studies, derive domain-specific dimensional weights for regulated-sector applications, and, in sequence, develop the agentic AI verification methodology. The sequencing principle is fixed: current framework first, domain-specific extensions next, agentic methodology later. Completion before extension. This function is currently held by the CEO. You will take ownership of execution while the CEO remains the framework author and final methodological authority during the founding-stage alignment period.

What you will do

• Study execution. Run pre-registered studies end to end: experimental design, balanced scenario matrices, data collection, and analysis (MANOVA, TOST equivalence testing, effect sizes, EFA/CFA where specified). Deliver manuscript-ready tables, figures, and reproducible analysis records.
• Instrument stewardship. Maintain MAAC instrument versioning, scoring protocols, and the derivation trace for all statistical parameters. Every parameter used in a certification decision must be pre-specified and documented; nothing is set by judgment after data collection begins.
• Criterion validity program. Own the research path that closes the criterion validity gap. Until that evidence exists, enforce the reporting-language guardrails: claims are scoped to process rigor and dimensional differentiation, never to outcome prediction, model rankings, or deployment suitability.
• Domain weight derivation. Lead the research deriving dimensional weights calibrated to regulated deployment contexts (healthcare, financial services, legal), producing defensible, documented weighting evidence for certification use.
• Scenario science. Advance complexity-validated scenario generation methodology, including LLM-assisted generation with independently verified cognitive demand properties.
• Publication pipeline. Manage journal submissions, reviewer responses, and revisions. Co-author with academic advisors where appropriate. Preprints under review are described as under journal review, never as peer-reviewed or published, until acceptance.
• Certification interface. Ensure the research program directly feeds the certification methodology: baseline reference distributions, floor thresholds, equivalence bounds, and drift-trigger definitions.

What we are looking for

• PhD required. Doctorate in a quantitative discipline: industrial and systems engineering, psychometrics, measurement science, quantitative psychology, statistics, or equivalent. The requirement is substantive, not cosmetic: the role demands credibility in academic and regulatory rooms, publication capacity, and defensible analysis under scrutiny.
• Psychometric depth. Demonstrated command of instrument validation: reliability, construct and criterion validity, factor analysis, and pre-registration discipline.
• Statistical fluency. Hands-on multivariate analysis in R, Python, or SPSS, with the ability to produce and defend a complete statistical chain of custody.
• Publication record. Peer-reviewed publications as a primary author, with evidence of carrying manuscripts through review.
• AI evaluation experience. Prior work evaluating large language models or AI systems is strongly preferred, including familiarity with the gap between benchmark performance and operational behavior.
• Claim discipline. A visible track record of claiming only what the evidence supports. Candidates who inflate findings, in their own work or in the interview, are disqualified.