AI × Web3 · Quant · ZK / DePIN
I build verifiable on-chain systems, AI agent applications, and quant tooling — fast iteration, clean architecture, reproducible setups.
Featured Work
Batch verification demo — storage layout for parallel execution, no write hot-spots. Batch verify 10 / 50 / 100 proofs per tx; metrics dashboard (gas, latency, success rate).
Solidity 0.8.24 · Hardhat · Next.js 14 · ethers v6 · Tailwind · PNPM monorepo
→ zkhahahaMulti-agent simulation — system prompts, personas, private memory isolation. Single API, multi-role agents; voting / night actions pipeline.
Agent orchestration · prompt protocols · state management
→ AI-llm-werewolf-arena-Quant track
| Area | Focus |
|---|---|
| Research / Modeling | Regime awareness, indicator diagnostics, signal design with leakage awareness |
| Backtesting | Crypto pipelines (Binance), MAE/MFE, win-rate vs tail-risk, stop-loss sensitivity |
| Execution | Bot-driven workflow on Linux/VPS, Docker, reproducible runs |
| Risk / Process | Position sizing, failure-mode thinking, robustness over “perfect params” |
Web3 / AI systems
Tech stack
More repos
Internship fit
Roles in Web3 engineering (protocol / infra / tooling), AI agent applications (orchestration, state, productization), Quant dev / data pipelines (research → backtest → execution).
Teams that value clean code, measurable iteration, and systems thinking.