Live web app · active developmentIndependent product

TutaLearn

AI-supported learning for Kenya's Competency-Based Curriculum (CBC)

Key result: Every CBC gradeSyllabus, quizzes and exams in place · grade-aware AI tutor

A mobile-first learning platform for Kenya's CBC: grade-by-grade syllabus content, quizzes and exams, progress tracking and report cards, with an AI tutor that knows the learner's grade and subject. A content pipeline validates generated material and keeps it in draft until it is reviewed and published. WhatsApp and M-Pesa features are coming soon.

Live app Source private

At a glance

Role: Full-stack & AI engineer
Status: Live web app · active development
Type of work: Independent product
Categories: Applied AI & Agents · Web & SaaS · Civic & Social Impact
Source code: Source private

Engineering highlights

Grade syllabus content, quizzes and exams in place, with progress tracking, report cards and parent views
Grade- and subject-aware AI tutoring
Content pipeline with validation and draft review before publishing; automated tests and CI/CD

Technology

Problem

Learners and parents following Kenya's CBC need curriculum-aligned practice and clear progress reporting. Generic AI tutors do not know which grade, subject or strand a learner is working on.

Who it’s for

CBC learners and their parents or guardians.

What I built

Syllabus content for each CBC grade, with quizzes and exams (completed, owner-reported).
Learner progress tracking, report cards and parent-facing views.
AI tutoring that uses the learner's grade and subject as context.
A curriculum operations pipeline: generated content is validated, held as a draft, then reviewed and published.
Supabase Auth with PostgreSQL and Row Level Security for learner data.
Integration architecture for WhatsApp and Safaricom Daraja (M-Pesa), ahead of launching those channels.

Architecture

Content pipelineGenerate → validate → draft → review → publish.
DataSupabase Auth and PostgreSQL with Row Level Security.
Learning appNext.js web app for lessons, assessments and progress.
TutorLLM tutoring with grade and subject context.
ReportingReport cards and parent views built from progress data.

Engineering challenges

Quality control for generated content

Generated lessons go through validation and stay in draft until they are reviewed and published.

Context-aware tutoring

The tutor is scoped to the learner's grade and subject, so answers match what the learner is studying.

Quality and trust

Automated tests and a CI/CD pipeline are in place (confirmed by the owner; test counts are not published here).
Row Level Security on learner data.

Limitations

WhatsApp and M-Pesa (Daraja) features are coming soon and are not live yet.
In active development; offline use is limited.