Exam-readiness engine · WAEC & JAMB · Lagos, Nigeria
Aaivanta — exam-readiness for Nigerian students
Aaivanta is an exam-readiness engine for Nigerian students preparing for WAEC and JAMB: it turns past-paper practice into adaptive questions, marks each answer with the reasoning a student missed, and gives a per-subject readiness score that names the exact topics to fix before exam day.
For students, and for the exam-prep centres and secondary schools that teach them.
Three parts. One engine.
Practice that adapts, marking that explains, and a readiness score that tells a student exactly what to fix next.
-
01
Adaptive practice
Questions pick themselves to the student's level, subject by subject — English, Mathematics, Physics, Chemistry, Biology, Economics.
-
02
Learned scorer in development
Explained marking
Every answer is marked with the reasoning the student missed, not just right or wrong. Objective questions are marked today; a learned scorer for short answers and essays is in development.
-
03
Readiness score
A per-subject score from 0 to 100, tracked against exam standard, that names the exact topics to fix next.
Try the practice engine.
Ten objective questions across Mathematics, Physics and English. Difficulty follows your running performance, every answer is marked with the reasoning behind it, and each subject's readiness score updates as you go. No account, no network — your progress stays in this browser.
Prototype today. Learned scorer next.
Today: the practice-and-marking engine runs in the browser — adaptive questions, rubric marking and a readiness score, all on the device. In development: a learned scorer that marks short answers and explains the mark.
- Adaptive questions
- Rubric marking
- Readiness score
- All on the device
- A learned scorer for short answers
- Marks the answer and explains the mark
From rubric prototype to learned scorer.
We do not run on NVIDIA hardware yet. As we move from a transparent, rubric-based prototype to a learned scorer, our build targets NVIDIA NeMo with CUDA to train the grading model, and NVIDIA Triton Inference Server with TensorRT-LLM — served as NVIDIA NIM — to mark a full exam sitting inside the marking window.
- Current
- practice and marking prototype — a static web app, rubric-based, running in the browser with data on the device; no NVIDIA hardware in use.
- In development
- a learned free-response scorer that explains the mark.
- Planned
- NeMo with CUDA to train the grading model; Triton with TensorRT-LLM (NIM) to serve marking at exam-season burst; RAPIDS for item analysis.
One founder, end to end.
Founded and built by Uthman Ajibola Sikiru, Founder.
Talk to us.
uthman@aaivantalabs.comStudents, exam-prep centres and secondary schools — write to us directly. Lagos, Nigeria.