[ /ABOUT.EXE ]

yo. i'm 0xUgly. this is my CYPHER TERMINAL.

i run Rupture Labs — a web3 gaming studio where we're trying to make blockchain games that don't completely suck (big mood).

i also fuck around with:

  • 🤖 AI/ML bullshit — fine-tuning models, running shit locally, RAG bots that answer my dumb questions, peptide chemistry models (lol)
  • 🎮 Web3 games — actually built a whole game engine (G3Engine) with Solana + EVM. send help.
  • 📱 Mobile apps — iOS (SwiftUI) + Android (Compose). mostly health apps with AI bolted on.
  • 🔗 Blockchain degeneracy — smart contracts, pump.fun integrations, NFT minting, token gating. ya know, the usual.

"i have no idea what i'm doing but it does seem to be working" — not every dev ever, including me

[ STATS.DAT ]
GitHub Repos 99 (i have a problem, send interventions) Commit Style 5% coding, 95% "git commit -m 'fix'"
Primary Stack Python, JS/TS, Swift, Kotlin Debug Time approx. forever
Coffee/Week ∞ Sleep Schedule "what's that?"
[ PROJECTS.SYS ]

🎮 G3Engine [SHIPPING]

Next.js 16 + Three.js + Solana Multi-Chain Game Engine (because existing ones suck)

literally built a full game engine from scratch. has 7 templates (puzzle, arcade, .io, racing, card, strategy, idle), pump.fun integration, NFT minting, token-gating, 2D visual editor with physics. code is probably trash but it works. ship it.

Stack: Next.js16 Three.js Solana Zustand Web3

"one day people will actually use this" — said every indie dev ever, forever

🤖 RuptureLabs-RAG [PRODUCTION]

FastAPI + Next.js + Ollama — $380/mo @ 600 users (zero pilot costs)

built this because our team kept answering the same questions on Discord. runs on our own servers, uses only local LLMs (Ollama), has chat UI + admin dashboard. 6-week rollout where i changed my mind 17 times. production means "i finally stopped coding and just shipped it."

Stack: FastAPI Next.js14 PostgreSQL pgvector Ollama Docker

"the bot is smarter than some users" — not even sorry

🧬 qwen3-peptide-ml [EXPERIMENTAL]

Fine-tuning Qwen3.5-0.8B-Base on 651 peptide/steroid data points

this one's for biotech/health — fine-tuning a small model on chemical/peptide data. runs on my MacBook Air M4 (16GB), takes ~15-20 min for 3 epochs using Axolotl. end goal: fully offline iPhone app via GGUF + llama.cpp (running LLMs on phones = peak degen).

Stack: Axolotl LoRA PyTorch GGUF llama.cpp

"molecular ML on a phone? yeah, i'm that guy now"

📱 HealthApp [BUILDING]

SwiftUI iOS app with multiple local AI backends + graceful degradation

started as a health tracker, then went full degensville and added AI. now supports Perplexity Sonar Debug + local Gemma4 architecture. graceful degradation pattern so it works even when shit breaks. testing on iPhone 17 Pro simulator (lol, not paying for real devices).

Stack: SwiftUI iOS Kotlin Compose

"health + AI = ???. i'm just here for the vibes"

[ TOOLBELT.SYS ]
▸ LANGUAGES
▸ Python (obviously)
▸ JS/TypeScript
▸ Swift
▸ Kotlin
▸ Solidity (lol) 		▸ FRAMEWORKS
▸ Next.js / React
▸ FastAPI
▸ SwiftUI
▸ Jetpack Compose
▸ Three.js (when i'm feeling fancy)
▸ BLOCKCHAIN
▸ Solana
▸ EVM chains
▸ Pump.fun
▸ Token bonding curves
▸ NFT minting 		▸ AI/ML
▸ Fine-tuning (Axolotl)
▸ Local LLMs (llama.cpp/Ollama)
▸ RAG / embeddings
▸ GGUF quantization
[ COMMS.LINK ]

▸ GitHub ▸ X / Twitter ▸ Email

"hit me up if you got a web3 game idea or some AI bullshit. or just wanna degens together."

⚠️ SYSTEM MSG: This website is running on 0 hours of sleep and approximately 420% CPU usage.
All errors have been swept under the rug. You're welcome.