Virtual Memory: Privacy-First Face Recognition App for Dementia Patients

#Healthcare

#Dementia

Event

LA Hacks

Team

Rikhil Rao, Arjun Chatha, David Samy, Akshaj Bharadwaj

Virtual Memory is a privacy-first iOS app built at LA Hacks 2026 that uses facial recognition and conversation tracking to help people with dementia remember who they are talking to. Built by Rikhil Rao, Arjun Chatha, David Samy and Akshaj Bharadwaj, the project was born from a deeply personal place: one team member's grandfather in India was living with dementia and struggling to recognize loved ones.

What it solves

Dementia affects millions of people worldwide and one of its most heartbreaking symptoms is the inability to recognize family members and friends during conversations. Existing tools either require constant manual input from caregivers or send sensitive personal data to cloud servers. Virtual Memory is built for low-connectivity environments and keeps all data on the device, making it accessible and private for vulnerable users.

What it does

Virtual Memory passively listens to natural conversations and automatically identifies what information matters. It summarizes key details from conversations, stores them in a personal memory wiki, and uses facial recognition to provide real-time context about the people the patient is interacting with. No manual note-taking required from the patient or caregiver.

How it works

Virtual Memory is built as a native iOS app in Swift using Xcode. It uses ZETIC Melange SDK to run speech summarization and facial recognition entirely on-device, ensuring the app works reliably even without internet access. All AI inference runs locally with no data sent to the cloud, keeping sensitive personal and medical information completely private.