🧠 Generative AI Assistant

🧠 Generative AI Assistant
By Vinay Sai Rajapu and Jaswanth Pavuluri
🌟 Project Showcase
AI Question Answering
Welcome to our Generative AI Assistant project! This AI system helps users ask questions related to Generative AI topics such as Transformers, Embeddings, Retrieval-Augmented Generation (RAG), GANs, and more.
Technology Stack
It uses OpenAI embeddings, Pinecone vector search, and a Streamlit frontend to create an interactive knowledge-based assistant.
🚀 Interactive Demo
Try It Yourself
👉https://jashu284-new-ai-app-cfl4fp.streamlit.app/
Ask AI Questions
You can ask questions about Generative AI, and the system will dynamically retrieve context from a knowledge base and answer them!
🛠️ Key Features
OpenAI Embeddings
Generates semantic vector representations of queries.
Pinecone Vector Database
Stores and retrieves top-matching knowledge base chunks.
Contextual Prompting
Builds a custom prompt based on retrieved context.
Streamlit Interface
Provides a clean, user-friendly chat experience.
Knowledge Base Upload
Users can upload their own .txt or .md files to customize the assistant.
Dark Mode Support
Users can toggle between light and dark themes.
Chat History
Maintains history of user and assistant messages for better conversation flow.
📂 GitHub Repository
Repository Link
You can access the repository at: https://github.com/Jashu284/NEW-AI
Repository Contents
Full source code (https://github.com/Jashu284/NEW-AI/blob/main/app.py)
Requirements file (requirements.txt)
ReadMe documentation
Setup instructions
Example outputs and screenshots
📄 Project Documentation
System Architecture
👉 https://github.com/Jashu284/NEW-AI/blob/main/Generative%20AI_Documentation.pdf
Implementation Details
The documentation includes implementation specifics and technical details.
Challenges Faced
Discussion of obstacles encountered during development.
Future Work
Suggestions for expanding and improving the system.
Ethical Considerations
Analysis of ethical implications and safeguards.
🏗️ Architecture Overview
User Input
Embedding Generation (OpenAI)
Similarity Search (Pinecone)
Context Formation
Prompt Building
Answer Generation (OpenAI Chat)
Display via Streamlit UI
We have also included architecture diagrams and prompt engineering examples below.
🎯 Future Improvements
Expand knowledge base to support multiple domains
Integrate multimodal AI (text + images)
Optimize query matching with re-ranking models
Add authentication for private knowledge bases
Improve user personalization with memory modules
🛡️ Ethical Considerations
Domain-Specific Answers
Ensured the assistant answers only Generative AI related queries.
Trusted Sources
Prevents misinformation by sticking to trusted knowledge sources.
Clear Disclaimers
Clear disclaimers added for unsupported queries.
Privacy Protection
No user data is stored to maintain privacy.
✅ Thank you for exploring our project!
Test
Try out our demo
Explore
Dive into our code
Contribute
Help us improve
Feel free to test, explore, and contribute!
Made with