Yavuz Uyar
I build AI systems grounded in how people actually learn.
Now
At Alpha School, I work on several projects toward one goal: making one-on-one teaching possible for every K-12 student, in every subject.
Selected Work
Bridge.
I built the integration system for third-party learning apps. It reads student activity, normalizes it into one measure of progress, and assigns courses and targeted lessons back into apps that rarely provide usable APIs. The feedback loop stays under three minutes.
TimeBack.
I work on systems that determine the best thing for each student to do next: the next course when they are ready, or a targeted lesson based on what they missed when they are not.
Invariant.
I built an AI browser agent that goes through learning apps like a student. It can try to solve lessons or cheat, including listening to browser audio and using microphone input. It produces an evidence-backed video report and grades the app against instructional invariants developed by learning scientist Carl Hendrick.
The Learning Standard.
I built it because most educational apps are evaluated for district procurement, not from a parent's perspective. It rates whether apps actually teach and whether families can see and control their data. Its automated news and research pipeline covers edtech and AI through a parent and learning-science lens.
TimeBack Insights.
An alarm outside the reward loop. It analyzes support issues, alerts, and student progress to find visible failures and students who are silently stuck, then surfaces patterns for remediation and product improvements.