rufatronics/Africa-rag-project1

Africa RAG Assistant using Gemini + Hugging Face Datasets (ReadyTensor Project 1) # 🌍 Africa RAG Assistant — ReadyTensor Project 1 A fully cloud-deployable Retrieval-Augmented Generation (RAG) assistant powered by **Gemini API** and **Hugging Face African datasets**. ## 🚀 Features - Automatically loads 3 African datasets (NER, stories, news) - Uses ChromaDB for contextual retrieval - Responds using Gemini 1.5 Flash via LangChain - Streamlit UI (upload optional, ask questions directly) ## 🧩 File Structure africa-rag-project1/ ├── app.py ├── src/ │ └── vectordb.py ├── requirements.txt ├── .gitignore └── README.md ## ⚙️ Setup (Cloud-Only) 1. Create a GitHub repo (public). 2. Add all files above. 3. Deploy to Streamlit Cloud 4. In **Settings → Secrets**, add: GOOGLE_API_KEY = "your_gemini_api_key_here" 5. Click **Deploy App**. ## 💡 Example Prompts - “Explain ECOWAS’s peacekeeping mission in West Africa.” - “Summarize renewable projects in Kenya.” - “Tell me a folktale from AfricanStories dataset.” ## 🧱 Datasets - MasakhaneNER - AfricanStories - Africa-News-Dataset