Decentralized File Storage Manager AI Agent

• 

  database uploaded with ai-extracted features

• 

  AI agent processing files

• 

  drag and drop window

Presentation

DFS_manager is an AI agent aiming at managing the storage of files in a decentralized database such as IPFS.

Using AI for managing databases is of crucial matter, as it's commonly known that decentralized storage might be slower at retrieval than a classic centralized database. Yet as web3 enthusiasts and concerned individuals about privacy and being resistant to hacking or censorship, we are still tempted to use decentralized storage against centralized solutions.

At term the DFS_manager might bridge the performance gap by optimizing the organization of decentralized databases for fast retrieval by your web3 app.

How I build it

After some research I found pre-trained model for detecting the tempo of a song dating from two year ago: https://github.com/pnlong/determine_tempo.

I adapted it to publish a compatible version on my hugging face profile: https://huggingface.co/jcarbonnell/tempo_detection.

I created a script in my app to call this huggingface model, and updated the app frontend with a drag and drop window, so that the app owner can easily upload more private content on its platform: https://github.com/partagexyz/1000fans.

What's next for Decentralized File Storage Manager AI Agent

In this first release, the whole process is already covered: from dropping audio and video files to the frontend interface, until the storage in the database, serving the app's content. The next steps will consist in complexifying the feature extraction with multiple AI models in the background, and making the agent compatible with more storage options.

Built With

• javascript
• nearai
• python
• react