Aegis: From Storage to Story

Inspiration

As time passes, countless digital photos end up forgotten in cloud storage, hard drives, and old devices — disconnected from the stories they represent. We wanted to build something that could protect, rediscover, and preserve those memories. Inspired by the idea of “legacy”, Aegis reimagines photo management through artificial intelligence, transforming raw storage into meaningful stories.

What it does

Aegis uses AI to analyse photos uploaded by users and automatically generate descriptive captions, contextual tags, and meaningful groupings. The web application provides Google Drive and OneDrive integration to facilitate upload of images from different sources.

It provides an intuitive review interface where users can:

Browse images using arrow navigation.
Edit titles and descriptions suggested by the AI.
Add, modify, or remove tags.

Apart from photo analysis, Aegis also performs image duplication detection to improve the efficiency of the workflow before data is sent to LLM.

In essence, Aegis helps users transform their old photos into meaningful stories.

How we built it

Frontend: React + TypeScript, styled with Material UI for a clean, modern interface.
Backend / AI Layer: Integrated with an LLM for contextual labelling and description generation.
Storage & Data Handling: Images and metadata are processed and organised via a structured mapping system, allowing for efficient preview and editing.
Design: Timeless and minimal design for easy user interaction.

Challenges we ran into

Ensuring smooth image handling and memory-efficient previews.
Keeping AI-generated descriptions relevant and accurate to the image context.
Designing an interface that balances automation with user control.
Handling diverse file formats and metadata consistency.
Integrating Gemini, frontend, and backend

Accomplishments that we’re proud of

Built a functional prototype that seamlessly merges AI labelling with user interaction.
Created a dynamic review interface inspired by professional photo curation tools.
Developed an end-to-end flow from upload → analysis → review → curation.

What we learned

How to effectively use LLMs for contextual image description and metadata enrichment.
The effect that good UI can have on user experience.
The importance of balancing automation with human oversight in creative tools.
User experience matters most when dealing with emotionally resonant data — such as personal memories.