SiblingSync is an on-device research platform that transforms existing family photo archives into identity-differentiated, de-identified longitudinal variables — without transmitting raw family photos for central analysis.
Developmental family science relies heavily on surveys, interviews, laboratory tasks, and structured observation. These methods are rigorous and important, but they are episodic, burdensome, costly to scale, and often limited to short observation windows.
Family photo archives are an underused ecological record. Families routinely accumulate longitudinal libraries that capture naturally occurring contexts, family groupings, sibling co-presence, and caregiver participation across years of ordinary life.
SiblingSync is being developed to test whether that record can support scalable, low-burden research variables while preserving family privacy.
Photo archives are selective records, not direct measures of family behavior.
Archives are shaped by caregiver habits, family routines, device access, and cultural practices around photographing children. SiblingSync therefore treats photo-derived variables as candidate indicators that must be evaluated against established measures — never as validated measures of family functioning.
SiblingSync does not infer parenting quality, parental intent, favoritism, fairness, differential treatment, clinical functioning, or global sibling-relationship quality from photographs.
SiblingSync is a privacy-preserving mobile research platform. It processes existing family photo libraries on-device, supports parent verification of persistent child identities, generates processing-quality metrics, and exports de-identified metric summaries for research analysis.
Existing family photo archives, analyzed within a participant-authorized window.
Identity-differentiated observations: which verified individuals appear, together, when, and in what contexts.
Identities are parent-verified, not inferred without human confirmation.
Standardized de-identified metric exports plus quality-control variables.
SiblingSync computes descriptive quantities from parent-verified identities across a defined analysis window. Candidate variables include measures of who appears together and how family life is documented over time. Two examples:
Quantifies how frequently identified siblings appear together across a defined analysis window, and how that changes over time.
A descriptive count of co-appearance. It is not a measure of sibling warmth or relationship quality.
Quantifies caregiver-child co-occurrence within documented photographic moments across the archive.
A descriptive count of co-presence. It is not a measure of caregiver involvement.
All photo-derived variables are treated as candidate research variables under empirical evaluation, not validated measures. The full variable set and its evaluation methodology are part of the research protocol rather than public documentation. Researchers interested in the specifics are welcome to get in touch.
Schematic of the kind of time series a research export would contain.
Values are synthetic and shown only to illustrate the structure of an identity-differentiated longitudinal export. They are not measurements, results, or study data.
Analysis, identity resolution and parent verification all happen on the participant's device. Only de-identified variables leave it.
The participant authorizes analysis of an existing photo library. Nothing new has to be captured.
Face detection, image processing and identity resolution run locally on the participant's device.
Parents confirm or correct identity mappings. Identities are never inferred without human verification.
Standardized variables are exported for analysis without transferring raw photographs or face-derived biometric representations.
Everything left of the double bar remains on the participant device.
A feasibility study of whether this approach can produce research-grade variables in real family archives, and whether those variables behave well enough to justify larger-scale validation.
Whether existing photo archives can be processed reliably across real participant devices and archive conditions.
Whether identity-differentiated variables can be generated reliably across real-world family photo archives.
Whether parents can complete identity verification within an acceptable workflow burden.
Whether standardized de-identified variables can be produced without raw-photo export. Privacy auditing is a must-pass criterion.
Whether candidate variables are stable enough, and relate to established family-science measures closely enough, to justify a larger validation study.
Whether participants find the workflow and its privacy model understandable and acceptable.
NICHD SBIR Phase I proposal in development. Maksin Technologies LLC is enrolled in the TEDCO SBIR/STTR Technical Assistance Program, Cohort 2026.
An operational on-device prototype exists and has been exercised in internal engineering runs. These are internal engineering results, not participant study data, and no study results are reported on this site.
SiblingSync does not assume that photographic patterns measure family relationships. Phase I will evaluate the reliability of selected photo-derived variables and their preliminary relationships with established family-science measures.
A photo-derived variable earns an interpretation only through that evidence. Until then it is a descriptive quantity computed from an archive — useful to study, not yet meaningful to report.
Preliminary associations would support larger-scale validation; they would not establish these variables as validated measures. Nothing on this site should be read as a validated measurement claim.
The architecture is built around data minimization, so that investigators, institutions and IRBs can evaluate exactly what does and does not leave a participant's device.
Original family photographs remain in the participant's device photo library and are not transmitted for central analysis.
Face detection and identity clustering are designed to run locally on the participant's hardware.
Face-derived representations are not exported as research data.
Parent verification of identity mappings occurs on the device, not in a central console.
Research datasets are designed to contain the de-identified metrics required for analysis, and data minimization is built into the workflow.
Participant consent, permission and child assent procedures are governed by the study protocol and IRB review. Participants control whether research export occurs.
For participating families, the plain-language version is simply that your family photos stay with your family. Full detail is in the Privacy Policy and Terms of Use.
SiblingSync does not store or back up original photos. If originals are deleted from the device photo library, they may be permanently lost. Keep independent backups.
SiblingSync is being developed as supplemental measurement infrastructure to sit alongside validated instruments — not to replace them. We are interested in conversations with investigators studying sibling and family processes, and in scientific input on measurement strategy and interpretation.
The following are potential research applications under development. None are validated applications today.
Families participating in SiblingSync research authorize on-device processing of an existing photo archive, verify family identities, complete study procedures as required by the protocol, and control whether de-identified research metrics are exported.
Which part of your existing photo library may be analyzed on your device.
Which identities are correct. Nothing is assigned to a child without your confirmation.
Whether de-identified research metrics are exported at all.
Join the research platform preview to help us test on-device processing, identity verification and the privacy-preserving workflow. We will email you from support@siblingsync.app with a build for your device or with next steps.
This is a research platform preview list, not enrollment in a research study. No NICHD-funded study recruitment is currently open. Any future research participation would occur separately, under an approved protocol with informed consent.
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Already testing SiblingSync? Share preview feedback
"I don't think we're taking as many photos of our third baby as we did with the first two."
That family observation started SiblingSync. With 4 kids under 4, our own archive grew faster than anyone could keep track of, and the first idea was a private way to organize it.
Development revealed a broader scientific opportunity. Family photo archives are large, longitudinal, ecological records of family life that researchers currently cannot analyze at scale while preserving family privacy. The hard and interesting problem was not organizing memories — it was whether an archive like that could yield defensible research variables at all, and what it would take to generate them without moving a single family photograph off the device.
SiblingSync evolved from a family-memory concept into privacy-preserving measurement infrastructure for developmental research. That evolution is why the scientific restraint on this site is deliberate: the interesting claim is not that photographs reveal families, but that some photo-derived indicators might be measurable, reliable, and worth evaluating.
— Abrham Tesfaw, Founder & CEO, Maksin Technologies LLC
NICHD SBIR Phase I proposal in development. Enrolled in the TEDCO SBIR/STTR Technical Assistance Program, Cohort 2026.