Ela Academic Assistant

A fully local Retrieval-Augmented Generation (RAG) application built with Microsoft Foundry Local, Streamlit, and SQLite for answering questions over academic documents.

The assistant performs semantic search on PDF and DOCX files, retrieves the most relevant document chunks, and generates grounded answers using a locally running language model.

Features

• Ask questions about academic PDF and DOCX documents
• Semantic search using vector embeddings
• Fully local AI inference with Microsoft Foundry Local
• Source-aware answers with document references
• SQLite vector database
• Interactive Streamlit interface
• Custom extractive parsers for structured university documents
• Works completely offline after setup

Technologies

• Python
• Microsoft Foundry Local SDK
• Streamlit
• SQLite
• NumPy
• PDFPlumber
• python-docx

System Architecture

User Question
      │
      ▼
Streamlit Interface
      │
      ▼
Semantic Retriever
      │
      ▼
SQLite Vector Database
      │
      ▼
Relevant Document Chunks
      │
      ▼
Extractive Parser
      │
      ▼
Phi-3.5 Mini (Foundry Local)
      │
      ▼
Final Answer

Example Questions

Here are some example questions the assistant can answer:

• What are the compulsory courses in the fifth semester of the Mathematics Department?
• Which documents should be prepared before an Erasmus internship?
• What information should be included in an Erasmus invitation letter?
• What are the eligibility requirements for TÜBİTAK 2209-A?
• Which Computer Engineering courses are exempt in the Double Major Program?

Project Structure

Ela-Academic-Assistant
│
├── app.py                 # Streamlit application
├── requirements.txt
├── README.md
├── LICENSE
│
├── documents/             # Source academic documents
├── data/                  # SQLite vector database
├── tests/                 # Debug and testing scripts
│
└── src/
    ├── answer_builder.py
    ├── chat_engine.py
    ├── chunker.py
    ├── database.py
    ├── document_loader.py
    ├── embedding.py
    └── retriever.py

Installation

1. Clone the repository

git clone https://github.com/YOUR_USERNAME/Ela-Academic-Assistant.git

2. Navigate to the project

cd Ela-Academic-Assistant

3. Create a virtual environment (optional)

python -m venv .venv

4. Activate the virtual environment

Windows

.venv\Scripts\activate

macOS / Linux

source .venv/bin/activate

5. Install dependencies

pip install -r requirements.txt

Usage

Run the Streamlit application:

streamlit run app.py

Then open your browser and navigate to:

http://localhost:8501

Screenshots

Home Page

---

Erasmus Internship Checklist

---

Mathematics Curriculum

---

Double Major Course Exemptions

---

TÜBİTAK 2209-A Information

Project Demo

A short presentation of Ela Academic Assistant, covering the project motivation, RAG workflow, development process, debugging experience, and key learnings.

▶️ Watch the Project Presentation

Future Improvements

• Hybrid search (Semantic + Keyword)
• Conversation memory
• OCR support for scanned documents
• Multi-document reasoning
• Re-ranking for more accurate retrieval
• Highlighting cited passages in documents
• Support for additional document formats

License

This project is licensed under the MIT License.

See the LICENSE file for more information.