Edge Runners 3.2

Start

elasticSearch

SirArthur7

Kabir Raj Singh

thelegtjet123

debasmit2020

Reminisce.AI is an comprehensive, multi-modal, AI application designed to enhance the lives of individuals with Dementia, Alzheimer's, different types of Amnesia, providing personalized memory support, query resolution, and routine management. Built to run efficiently on edge devices, the app leverages Llama 3.2-1B-Instruct for offline processing and Llama 3.2-11B-Vision for vision-driven tasks, allowing it to function both online and offline depending on connectivity. This is complemented by speech-to-text, text-to-speech models for smooth, natural language, audio-based interaction.

By integrating state-of-the-art SPLADE and CLIP embedding models, our app can understand both visual and textual data in a highly efficient manner, enabling responsive and context-aware interactions. The app uses a Knowledge Graph-based Retrieval-Augmented Generation (RAG) architecture, which allows it to store and retrieve detailed, personalized information, ensuring that users receive relevant and accurate responses based on their past interactions.

With these powerful AI models, Reminisce.AI helps users relive cherished memories through image slideshows, offers real-time assistance with everyday queries, and provides scheduled alerts for meals, medication, and sleep routines, ultimately improving both the quality of life for individuals with dementia and peace of mind for their caregivers.

Reminisce.AI is an AI app for individuals with dementia, providing memory support, query resolution, and routine management. It uses Llama 3.2 models for offline processing, SPLADE, CLIP embeddings, and RAG architecture for personalized interactions.

"elasticSearch team on Edge Runners 3.2 Hackathon"

Team Idea

Submission