Technology Stack

Backend Infrastructure

Supabase-Powered Services

  • Authentication :Supports email/password and Web3 wallet-based logins.

  • Database: Tracks Users, GPUs, Rentals, Models, Bots

Database Management

RPCs

  • increment_user_credits: Add GPU credits with $TRNV

  • get_training_status: Live training monitoring

UX/UI Stack

  • UI: Dark Mode + Shadcn/ui Kit + Lucide Icons

  • Animations: Framer Motion

  • Responsive: Mobile-first & accessible

  • Framework: TailwindCSS

  • UI Kit: shadcn/ui for standardized components.

  • Notifications: Toasts and modals for feedback.

  • Icons: Lucide React icons.

  • Skeletons: For loading states.

Technology Stack

Layer

Tech Stack / Tools

Description

Frontend

Next.js (React), TailwindCSS, ShadCN UI, Vercel

User interface built with React + Tailwind, deployed via Vercel. Supports SSR and fast UI updates.

Backend

Node.js, Express, Supabase

Handles logic, routes, auth, and database queries. Supabase for auth, storage, and DB services.

Authentication

Supabase Auth, WalletConnect, MetaMask, Magic.link

Web3 wallet login (MetaMask) or traditional login (email/password) via Supabase & WalletConnect.

Database

PostgreSQL (via Supabase)

Stores user profiles, models, GPU usage history, etc. Supabase provides realtime & GraphQL APIs.

File Storage

Supabase Storage, IPFS (optional)

Stores user-generated assets, AI models, and bot configurations.

GPU Node Layer

Mock GPU API, Real Cloud GPU APIs (e.g. RunPod, Banana.dev, LambdaLabs)

Users rent GPU power via mock APIs (for testing) or integrated providers via SDK or REST APIs.

AI Training

Custom Python SDK, Hugging Face Transformers, Stable Diffusion, Torch

Executes training jobs (diffusion, text, voice models) using backend pipelines triggered by frontend.

Bot Deployment

Telegram Bot API, Trainava Bot Manager (Custom)

Deployed models are converted into Telegram bots with minimal config. Managed and hosted by app.

Smart Contracts

Solidity, Ethereum, ethers.js, Hardhat

Handles rewards, staking, renting logic, and user ownership records via $TRNV token.

Analytics / Logs

Supabase Logs, PostHog, Custom Logging

Tracks user behavior, training usage, and bot activity.

CI/CD & Deployment

Vercel, GitHub Actions, Docker

Automated testing & deployment for both backend and frontend.

🔧 Breakdown of Components:

  1. Frontend / UI

    • Built with Next.js + TailwindCSS

    • Connects via Web3 wallets (MetaMask) or email/password (Supabase auth)

  2. API Layer

    • Handles routing, user interaction, logic validation, and middleware

    • Powers UI actions like training, renting GPUs, and deploying bots

  3. Authentication

    • Users log in via MetaMask (Web3) or Supabase email

    • User roles and session tokens managed via Supabase Auth

  4. GPU Marketplace System

    • Buy or rent GPU power

    • Connects to real or mocked GPU nodes (future integrations: RunPod, LambdaLabs)

  5. AI Training Engine

    • Pre-built model templates (e.g. Stable Diffusion, Whisper)

    • Custom training using GPU power

  6. Smart Contract Layer

    • $TRNV token interaction (rent, reward, stake)

    • Verified on Ethereum (future support: Arbitrum, Base)

  7. Database (Supabase)

    • Real-time syncing

    • Stores user profiles, training jobs, GPU listings, bot deployments

  8. Token Logic & Rewards

    • Token flow logic handled by smart contracts

    • Earn by providing GPU or deploying AI bots

  9. Telegram Deployment

    • Models can be deployed as Telegram bots

    • Bots managed via Telegram Bot API and database metadata

Last updated