const funboy = {
role: "Founder & CEO @ AIRewardrop",
origin: "gaming & retrogaming culture -> hardware -> IT/telco infrastructure",
now: "AI-native systems for crypto, social automation & trading intelligence",
belief: "AI agents should observe, reason, test and act, not just chat",
runsOn: ["self-hosted GPUs", "local LLMs", "real production infra"],
};I didn't start with AI. I started with machines. Years spent inside gaming and retrogaming communities, building and tuning hardware, taught me how digital products create attachment, how online economies work, and how technology actually behaves under load. That grew into founding and running companies in IT and telecommunications: networks, servers, security, real customers, real uptime.
Then came crypto and DeFi: token economies, on-chain data, market structure, community incentives. And finally the convergence point: AI agents that live inside those markets and communities, powered by infrastructure I design and host myself.
Everything I build converges on one thesis: moving AI from generated content to operational agency. Agents that analyze data, engage communities, assist decisions, and execute actions through secure, measurable workflows.
| 🤖 Embodied AI Agents 3D avatars with voice, vision and memory, on web, Telegram, Twitch and Unreal Engine |
🧠 Local LLM Infrastructure Self-hosted inference, multi-provider gateways, quota routing, no API-key dependency |
⛓️ On-chain Automation Minting bots, trading intelligence, PnL tracking: agents acting in real markets |
🔬 Frontier Research Sensor states injected directly into LLM latent space, beyond text tokenization |
AIRewardrop is my main venture: an AI-agent ecosystem for crypto, covering social interaction, DeFi intelligence, automated research, trading analysis and real-time community engagement.
| Project | What it does | Stack |
|---|---|---|
| 🎭 airifica-web | Per-user AI agent avatar: market opinions, trade setups, on-chain actions via natural language. Local LLM + TTS/STT + Web3 wallet | Vue ElizaOS Web3 |
| 📈 AIRTrack | Auto-trading tracking dApp: receives trades from the AIR³ agent and tracks PnL for social engagement | TypeScript |
| 🔀 GemRouterFE | OpenAI-compatible router for Gemini: multi-key routing, quota tracking, fallback, DeepSeek/Ollama-compatible surfaces, operator dashboard | TypeScript Node |
| 🎙️ GoonersBot | Group-native Telegram AI character: reads the room, remembers lore, voice notes, vision, web search. Self-hosted, uncensorable | TypeScript MongoDB |
🧠 LLM Infrastructure & Gateways
- NexGate: local proxy routing to ChatGPT, Claude, Gemini & Grok via real browser sessions. No API keys, automatic account rotation, live provider dashboard.
- GemRouterFE: OpenAI-compatible router for Gemini with multi-key routing, quota tracking and fallback.
- TeGem: Telegram bot driving the Gemini web UI directly through Playwright. Streaming replies, image generation, per-chat sessions.
- virtflux: image analysis, image generation and TTS behind one front end + API.
🔬 Research & Embodiment
- Somatic: SomaBridge injects continuous physical sensor states directly into LLM latent space through a C++/LibTorch projection layer, bypassing tokenization. Low-latency somatic-aware AI.
- Eliza2Face: the voice & face layer. AI text becomes speech and avatar animation for live agents in Unreal Engine / Audio2Face.
- AIR3-ElizaOS-UnrealE55-SDK: Unreal Engine plugin for conversational AI with environment perception and in-world action-taking.
⛓️ Web3 & Market Automation
🤖 Social Agents & Clients
- RIP2ETF: zero-hallucination ETF comparison agent with charts, delivered in Telegram & Discord.
- AIR3twitchBot: crypto quiz & engagement bot living inside Twitch chat.
- ElizaOS clients: Twitch · Reddit · Farcaster · Telegram-Airifica, extending agent presence across every platform that matters.
From simple AI-generated content to agents that observe, reason, test, communicate and operate inside real digital markets.
Systems that reduce operational friction, improve decision-making, automate complex workflows, and create new forms of interaction between users, markets, communities and autonomous digital agents. All running on infrastructure I control, end to end.
Let's build something that acts.
