Elamur Research⌕
☰
home / papers / 2026 / EduNaija AI Tutor: A Multi-Agent Retrieval-Augmented Generation System for Nigerian Curriculum Education

EduNaija AI Tutor: A Multi-Agent Retrieval-Augmented Generation System for Nigerian Curriculum Education

Paper · 2026 · Israel Olanrewaju Odeajo and Edifon Emmanuel Jimmy

Summary

Nigerian students preparing for national exams such as WAEC, NECO and JAMB often lack affordable, personalized tutoring. Private lessons cost roughly ₦5,000–₦50,000 per month and are concentrated in cities. General-purpose LLMs and existing tutors like Khanmigo are not built around local syllabi. The authors present EduNaija AI Tutor, a multi-agent RAG system. A central orchestrator identifies what the user wants and routes the request to specialized agents for conversational tutoring, question answering over documents, or generating practice assessments. Learners can upload their own study materials, which are parsed, chunked, embedded and indexed for retrieval, and LangChain coordinates the agents. The system supports English, Yoruba, Hausa and Igbo, although non-English input is translated into English for reasoning. In a preliminary evaluation, 50 professional evaluators from three departments assessed the tutor. The authors report that it can produce curriculum-aligned explanations and practice tests, but the extracted text gives no detailed scores. The work is an early attempt to ground LLM-based educational tools in an African national curriculum rather than a Western one.

Key points

  • The system targets the WAEC, NECO and JAMB exams. More than two million candidates register for JAMB each year.
  • The retrieval pipeline works as follows:
    • PyMuPDF extracts the text.
    • RecursiveCharacterTextSplitter splits it into chunks (size 1024, overlap 100).
    • OpenAI text-embedding-ada-002 produces the embeddings.
    • FAISS stores the vectors.
    • LangChain handles agent orchestration.
  • The architecture is hierarchical: a main orchestrator classifies intent and dispatches queries to task-specific agents.
  • The system covers English, Yoruba, Hausa and Igbo. Its main acknowledged limitation is the translation pivot to English, which may lose nuances of Yoruba or Igbo.
  • The evaluation involved N = 50 professional evaluators across three departments, and its results are described as preliminary. The tool is explicitly positioned as a supplement to schooling, not a replacement.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

Also on the byline: University of Benin.

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksInformation Systems · 1,437 worksMedicine · 241 worksHealth Informatics · 184 works

🧭 Topics 1Research subjects and languages the work is about.Information retrieval & search · 473 works

Cite

Cite as Israel Olanrewaju Odeajo and Edifon Emmanuel Jimmy. 2026. EduNaija AI Tutor: A Multi-Agent Retrieval-Augmented Generation System for Nigerian Curriculum Education. Proceedings of the 7th Workshop on African Natural Language Processing (AfricaNLP 2026).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.africanlp-main.11

@inproceedings{odeajo2026edunaija,
  title = {{EduNaija AI Tutor: A Multi-Agent Retrieval-Augmented Generation System for Nigerian Curriculum Education}},
  author = {Israel Olanrewaju Odeajo and Edifon Emmanuel Jimmy},
  booktitle = {Proceedings of the 7th Workshop on African Natural Language Processing (AfricaNLP 2026)},
  year = {2026},
  doi = {10.18653/v1/2026.africanlp-main.11},
  url = {https://doi.org/10.18653/v1/2026.africanlp-main.11}
}

References

6 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. David Ifeoluwa Adelani, Graham Neubig, Sebastian Ruder, Shruti Rijhwani, Michael Beukman, Chester Palen-Michel, Constantine Lignos, Jesujoba Alabi, Shamsuddeen Hassan Muhammad, Peter Nabende, and Cheikh M. B. Dione. 2022. MasakhaNER 2.0: Africa-centric transfer learning for named entity recognition.…
  2. Khan Academy. 2023. Khanmigo: AI-powered teaching assistant and tutor. Online resource. *Table 1: Demographics of the 50 evaluators across major academic departments.* Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau… find
  3. Criteria Yoruba Igbo Hausa Language Accuracy Edu. Info Accuracy Doc. Handling Output Structure 8.0 8.0 7.0 8.0 7.0 6.0 6.0 6.0 7.0 8.0 8.0 9.0 Normalized (%) 77.5% 62.5% 80.0% S. Liu and 1 others. 2025. Improving large language model applications in biomedicine with retrieval-augmented generation: A… find
  4. West African Examinations Council. 2022. Waec syllabus and regulations for west african senior school certificate examination. Technical report, WAEC, Lagos, Nigeria. \#\# Conclusion EduNaija demonstrates that Multi-Agent RAG systems can effectively bridge the educational gap in Nigeria. Future work w… find
  5. Beverly Park Woolf. 2010. Building Intelligent Interactive Tutors. Morgan Kaufmann. find
  6. World Bank. 2022. Nigeria poverty assessment 2022: A better future for all nigerians. Technical report, The World Bank Group. find

Updated

This page in Markdown → https://md.research.elamur.ai/free/en/papers/odeajo-2026-edunaija-ai-tutor-multi-agent/