Separate the signal before the referral.
Bilingual students who do not exit ELL status by Grade 3 are 156% more likely to be misclassified. XODA provides the diagnostic layer that standard systems were not built to supply. (Cruz & Firestone, 2022)
The math is still there.
It was never missing.
The tool that could see it — that was missing.
Standard assessments run mathematical ability and language fluency through the same lens. For a student still developing English, that lens measures both at once — and reports one score.
When a bilingual student in a Title I classroom writes their math reasoning in abbreviated, phonetic, or mixed-language form, a standard rubric sees incomplete work. The diagnostic system built here sees a different signal: the mathematical structure beneath the surface.
That separation is the design. It is not an accommodation. It is the architecture.
Bilingual students who do not exit ELL status by Grade 3 are 156% more likely to be misclassified. XODA provides the diagnostic layer that standard systems were not built to supply. (Cruz & Firestone, 2022)
Fifteen years in Title I classrooms. The pattern was the same every time: the math was always there. The students who write in consonant skeletons, who mix Spanish and English mid-sentence, who abbreviate every word — they are not behind in mathematics. They are ahead in code-switching.
XODA reads the math signal the same way an experienced teacher does. Architecturally. Without guessing.
If your child solves problems in their head, explains solutions in a mix of languages, or writes answers that look abbreviated — those are not gaps. They are signals. XODA is built to read them.
QuizMyBrainz was founded by two educators who brought over 30 years of combined Title I bilingual classroom experience across multiple North Texas school districts to a single diagnostic question. Every misclassification they witnessed — every bilingual student routed toward an intervention that tested language fluency instead of mathematical ability — sharpened the same question: why doesn't a tool exist that separates these two signals?
The answer was that building it required both classroom-level ground truth and systems-level architecture. That combination took over 30 years to accumulate and one focused research sprint to deploy.
Accepted Research · 19th International Conference on Educational Data Mining (EDM 2026).
2015 Fort Worth ISD Campus Teacher of the Year, recognized by Texas House Resolution H.R. 2522, 84th Legislature.
Peer-reviewed manuscript under preparation for BEA 2027 (ACL Workshop on Innovative Use of NLP for Building Educational Applications).
Raw student text never reaches any AI model.
That is not a policy commitment.
That is the architecture.
FERPA and HIPAA Safe Harbor compliant.
How we protect student data →