victorachede/offline-scholars

Fully offline AI tutor for JAMB, WAEC and NECO exam prep — Africa Deep Tech Challenge 2026 # Offline Scholars > A fully offline AI tutor for Nigerian students preparing for JAMB, WAEC, and NECO. > Built for the **Africa Deep Tech Challenge 2026**. --- ## The problem Millions of Nigerian students prepare for JAMB every year in areas with no reliable internet. Private tutors cost money most families don't have. Past question books go out of date. And when a student gets a question wrong, nobody explains *why*. Offline Scholars fixes this. It runs entirely on a standard laptop — no internet, no API fees, no subscription. Open it, study, pass. --- ## How it works Three modes, one install: ### 1. Practice Mode The AI presents real JAMB/WAEC/NECO past questions one at a time. Student answers. AI marks it and explains the answer in plain conversational language — not just "correct answer is B" but *why* B is correct, what concept it tests, and what JAMB typically asks about it. ### 2. Ask Anything Student types any question — "I don't understand equilibrium" or "explain the difference between speed and velocity" — and the local AI explains it from scratch. Like having a private tutor available 24/7 with no data required. ### 3. Mock CBT Exam Timed, same format as the real JAMB CBT. 40 questions, subjects mixed, countdown timer. Submits and shows full score breakdown — every wrong answer explained. --- ## Hardware requirements Designed for the contest constraints — and for the average Nigerian student's laptop: - **RAM:** 8GB (runs in ~4-5GB) - **GPU:** Not required — CPU only - **Storage:** ~2.3GB for models - **OS:** Windows 10+, macOS 12+, Ubuntu 20.04+ --- ## Quick start ```bash # 1. Clone git clone github.com cd offline-scholars # 2. Install Ollama (one time) # Download from ollama.com and install # 3. Install Python dependencies pip install -r requirements.txt # 4. Download models (one time, ~2.3GB, needs internet) python scripts/download_models.py # 5. Launch — now fully offline python app …