Skip to content
erzurumlu.dev
← All projects

Karnova.

Beyond a grade, toward a student’s needs.

Learning-objective analysis and adaptive content for individual learners.

Role
Backend & AI
Scope
3 people
Year
2026

Student analysis

◉   Ahmet Yılmaz
OverviewLearning outcomesRecommendations

Course progress

Turkish
0%
Mathematics
0%
Science
0%
Social Sciences
0%

Recommended lessons

YKS
FunctionsMathematics · Video · 12 min
Meaning in paragraphsTurkish · Practice · 15 questions
Force and motionScience · Activity · 10 min
Performance trendLast 7 weeks
Week 1Week 2Week 3Week 4Week 5Week 6Week 7
Illustrative interface · Sample data
The problem

A grade is not a diagnosis.

Make it clear which learning objectives a student needs help with.

Traditional assessment often only shows the result. Understanding which objectives a student struggles with makes it possible to offer the right content. Karnova makes this visible through data and AI.

Student answersQuiz and exam answers are analysed at objective level.
Objective analysisMap each response to learning objectives.
Content suggestionsSuggest content for learning gaps.
My contribution

Data model, REST API, Gemini integration and role-based access.

I focused on a secure, scalable and maintainable technical foundation.

Objective-based data model

Students, questions, curriculum objectives and answers in one model.

Function calling grounded in real records

Gemini answers are tied to actual student objective records.

Data isolation between schools

Multi-tenant architecture and role-based access control.

← Learning objectivesMathematics⌄
Ratios and Proportions0%
Ratio0%
Direct proportion0%
Inverse proportion0%
Recommended content
Ratios · Key conceptsVideo · 12 min
Direct proportion problemsPractice · 15 questions
Ratios in everyday lifeReading · 10 min
Outcome

Objective-level reporting and adaptive content recommendations.

Teachers can identify strengths and gaps more clearly, then suggest content that meets the student’s needs.

What I learned

“It stopped guessing and started querying.”

Function calling ties the model’s answers to actual student records. A reliable answer starts with a reliable data path.

Tech stack.

  • Django 6
  • Python
  • PostgreSQL
  • Gemini API
  • REST API
  • RBAC
Technical notes and project scope

The problem and my role

When a student scores 40 on a maths test, that number says nothing about where they got stuck. A teacher with thirty students cannot diagnose each one individually — and even if they could, another test arrives the next day. A grade is an outcome, not a diagnosis.

In a three-person team I own the backend and AI side: data model, REST API, Gemini integration and the authorisation layer. The interface and the operations, content and QA side belong to the other two.

Implementation decisions

  • I built the data model around learning objectives rather than grades: every question maps to one or more curriculum objectives, and answers aggregate at the objective level.
  • Instead of letting the AI assistant free-write, I gave it function calling — the model reaches the student’s real objective data through functions it invokes itself. That structurally prevents it from telling a thriving student they are failing.
  • Each school’s data is isolated: a multi-tenant setup with role-based access control gives teachers, students and administrators separate doors.

Outcome and learning

The platform now delivers objective-level reporting and adaptive content suggestions. The source is currently private; the architecture documentation is public.

What makes a language model trustworthy in production is not a better prompt but a data path it cannot fabricate. Free-form generation gives you no way to stop it from saying something false about a student; with function calling its answer is tied to the actual record in the database. It stopped guessing and started querying.