AI-Powered Assistant for User Signature

As a visitor to a Unisala user's signature page, I want to interact with an AI assistant that can intelligently answer questions about the user's professional background, expertise, and experiences, so that I can quickly understand their capabilities and knowledge areas without having to manually parse through their entire profile.

signature/Prashant-Basnet

Acceptance Criteria:

Core Features

• Must answer fundamental questions like:
• Basic profile information: "Who is [User]?"
• Knowledge assessment: "What topics has [User] studied most?"
• Expertise evaluation: "How experienced is [User] in [Topic]?"

Data Integration

• Successfully retrieves and processes:
• User profile details (tags, descriptions)
• Platform activity (notes, posts, contributions)
• Knowledge graph relationships
• Resumes / description provided by user

Response Quality

• Provides concise, relevant answers
• Includes supporting evidence from user's projects/posts
• Handles edge cases with appropriate fallback responses
• Maintains conversation context for follow-up questions

Technical Requirements

• Response time under 2 seconds
• Supports concurrent users across different signature pages
• Seamlessly integrates with Unisala's UI
• Uses lightweight model architecture (DistilBERT/similar)
• Deployable on Hugging Face or SageMaker cost-effectively

MVP Success Metrics

• Response accuracy above 85%
• Performance: Average response time below 2 seconds
• System uptime above 99%
• Leverage lightweight models and serverless infrastructure to minimize hosting costs.
• Deployment: Deploy on Hugging Face or Amazon SageMaker.
• Ease of Use: Visitors should interact through a simple UI embedded on Signature pages.

Technical Notes:

• Model Type: Start with lightweight pre-trained models (e.g., DistilBERT) with potential enhancements using auto-encoders for personalisation.
• Integration: Use API-based communication between the chatbot backend and the Unisala platform.
• Hosting Options: Explore Hugging Face Spaces for the MVP and AWS SageMaker for production scalability.

This story captures the essential goals and expectations for building an AI chatbot for Unisala's Signature pages.