← Back to home

M-P Wonderful Life

Wonderful Life is humanitarian. It collects free-text profiles of persons in need, creates embeddings, stores profiles & vectors on IPFS, stores CIDs on blockchain. Given a helper program, it makes its embedding and matches the closest profiles to help, based on the embeddings

Problem Statement

Machu Picchu is like the Inca citadel of same name, that is made of huge stone blocks tightly fit together. Machu Picchu is composed of tools in DeFi, IPFS, AI, Earth Observation, low cost Raspberry Pi-style SBCs with AI HATs. In this hackathon we will address a subset of Machu Picchu functions.The following resources explain Machu Picchu and its technologies: Our scenario: • the actors are the person in need who posts a free text profile : age, gender, location, occupation, needsan international helper organization, represented by an agent that creates an embedding of the profile (a vector of 512 parametres) and stores the profiles + embeddings on IPFS and their CID hashes on blockchain • an agent that would create in free text a Helper Program (objectives, target region, target group, duration, KPI, scale, budget), make an embedding of the Helper Programan agent that would match the vector of the persons in need and match with the vector of the Helper Program, finding the person who are the most entitled for help

Solution

Tech StackFrontend: Next.js 14, ethers.js, TailwindCSS,Framer Motion, etc.Authentication: Privy for wallet connection and social loginBlockchain: Remix IDE, Deployed on Sepolia testnet, etcStorage: IPFS via Pinata for decentralized storage of profiles and program dataAI/ML:Autonome (Eliza) AI Agent integrated with Telegram and Twitter to assist users with undestanding of application and answer queries related to it.Vector similarity matching for program-profile compatibilitySmart Contracts: Solidity for storing CIDs and managing program dataKey Features & ImplementationProfile Management:User profiles are converted to embeddings using AI APIBoth original profile and embeddings are stored on IPFSIPFS CIDs are stored on-chain for verificationHelper Program Matching:Program requirements are converted to embeddingsVector similarity comparison with stored profilesAutomatic matching of suitable candidatesDecentralized Architecture:No central database - all data stored on IPFS and blockchainSmart contracts manage data integrity and access controlFully decentralized authentication via PrivyNotable Technical ImplementationsCustom vector similarity algorithm for profile matchingHybrid storage approach combining IPFS and blockchainResponsive UI with server-side rendering for better performanceReal-time updates using blockchain eventsIntegrated with Telegram an TwitterIn-App Chatbot to assit users throughout the application.

Hackathon

Agentic Ethereum

2025

Contributors