✨ Hi, I'm Xinqiang Yu — building the future, one agent at a time

• 🎓 Cyberspace Security @ Tianjin University (2024 – 2028) · academic-year ranking 2/102 (weighted) / 1/102 (comprehensive)
• 🏆 Selected awards: ISCC Finals 🥇 · Huawei Software Elite Challenge (regional) 🥇 · Huawei ICT Ascend AI (national) 🥈 · 4C (national) 🥉

🔬 Research directions — full details on my academic homepage:

• 🤖 AI for AI — automating AI research with AI: AutoSFT, a multi-agent harness that automates SFT data research by iteratively rewriting the training data recipe with the training stack frozen; and equiNAS, a typed DSL for certified evolution of equivariant architectures, where every LLM-proposed mutation is machine-checked for type safety, equivariance, and dimension consistency before any GPU time is spent.

• 🛡️ LLM Safety — post-training safety alignment of large language models: safety-aware preference optimization, personalized safety adaptation, and multi-turn jailbreak robustness.

• 🕸️ Multi-Agent Systems — multi-agent collaboration for security and competition: adversarial multi-agent scoring (4C), multi-agent RL (NeurIPS Lux AI Season 3, Kaggle bronze), and an AutoHunter-style multi-agent system for automated SRC vulnerability discovery, with humans only as final reviewers.

• 🌱 Learning playlist: MIT Missing Semester (CN) · NTU ML 2025 (Hung-yi Lee) · MIT 6.1600 Security

• 📌 Now: assembling the SRC bug-hunting agent crew — target collectors, recon workers, exploitability reviewers, and a hunter that goes for mass exploitation.

• 🏫 Visiting Student · Zhongguancun Academy, Beijing (北京中关村学院) — Jul 2026 – Sep 2026
• 🔬 Research Assistant · Tsinghua Shenzhen International Graduate School (清华大学深圳国际研究生院) — May 2026 – Present
• 📡 Research Intern (online) · China Mobile — personalized LLM safety & trustworthy alignment — concluded Mar 2026

Generated nightly via GitHub Actions — because even commits deserve some neon art.