Education
University of Southern California — M.S., Computer Engineering (Honors Program)
2025 – 2027
Fortis Lab · Trustworthy & self-evolving AI agents · Los Angeles, USA
Shanghai Jiao Tong University — B.Eng., Electrical & Computer Engineering
2022 – 2026
Minor in Computer Science · Shanghai, China
Selected Publications (2026 · first-author unless noted)
- A. Yuan, Z. Su, Y. Zhao. “AEGIS: No Tool Call Left Unchecked — A Pre-Execution Firewall and Audit Layer for AI Agents.” arXiv:2603.12621 · Cited by 7 · 371★ open-source.
- P. Z. Wang, A. Yuan, H. Zhang, X. Hu, Y. Zhao, S. Jiang. “WeClawArena: An Auditable Sandbox and Benchmark for Cross-User Agent Collaboration and Security in Human-Centered Agent Networks.” EMNLP 2026 · arXiv:2608.03499.
- Y. Nian, A. Yuan, H. Zhang, J. Li, Y. Zhao. “Auditable Agents.” KnowFM @ ACL 2026 · ACM AI Summit 2026 · Cited by 9.
- X. Zhou, A. Yuan, et al. “When Simulation Lies: A Sim-to-Real Benchmark and Domain-Randomized RL Recipe for Tool-Use Agents.” NeurIPS 2026 · FAGEN @ ICML 2026 · Cited by 2.
- A. Yuan, Y. Nian, H. Zhang, Z. Su, Y. Zhao. “SEVA: Self-Evolving Verification Agent with Process Reward for Fact Attribution.” AI4GOOD @ ICML 2026.
- A. Yuan, H. Zhang, Z. Wang, Y. Zhao. “Sovereign-OS: A Charter-Governed Operating System for Autonomous AI Agents with Verifiable Fiscal Discipline.” arXiv:2603.14011 · 102★ open-source.
- A. Yuan, H. Zhang, S. Nazarian. “AgentIR: A Workload-Adaptive Cascade Retrieval Substrate for Long-Term Conversational Memory.” arXiv:2605.25092.
- A. Yuan, T. Shen, D. Zhang. “Not All Thoughts Need HBM: Semantics-Aware Memory Hierarchy for LLM Reasoning.” arXiv:2605.09490.
Experience
Topify.AI — Forward Deployed Engineer
Nov 2025 – Jan 2026
- Designed & deployed an autonomous GEO/SEO agent system (LangGraph, Inngest, Claude API) orchestrating 6 sub-agents into a closed-loop pipeline that tracks brand share across AI search engines (Perplexity, ChatGPT, Gemini) and auto-tunes content strategy.
- Cut content-ops headcount 80% and drove 200% revenue growth to $3M ARR, replacing a 5-person team with one self-optimizing workflow.
JD.com — Recommendation Algorithm Engineer
Dec 2024 – Feb 2025 · Beijing
- Built an end-to-end generative recommendation framework (semantic-ID retrieval) unifying candidate generation & ranking; used Qwen-VL + NER to mine 18 attributes and improve codebook utilization for ID quantization.
- Designed a multi-view user module fusing short/long-term behavior via QFormer aggregation of hierarchical semantic embeddings.
Optiver — Algorithm Engineer
Dec 2023 – Feb 2024 · Shanghai
- Ported an LSTM volatility model from Python to C++/LibTorch with custom CUDA kernels; 70% latency reduction (5 ms → 1.5 ms) via TensorRT quantization & kernel fusion for automated market-making.
- Built a real-time tick-level order-book ingestion module in C++/ZeroMQ for Chinese commodity futures exchanges.
Enjoymi Games (欢动科技) — Software Engineer
Aug 2023 – Oct 2023 · Shanghai
- Maintained core mobile-game SDKs (login, payment, platform integration); architected a unified auth interface over the HMS Account SDK and custom Android channel packages (Xiaomi).
Flagship Open-Source & Ventures
- AEGIS (371★, TypeScript) — Runtime policy enforcement for AI agents: cryptographic audit trail, human-in-the-loop approvals, kill switch. Zero code changes. → aegistraces.com
- Sovereign-OS (102★, Python) — Charter-governed AI orchestration: CEO plans → CFO approves → Ledger tracks every cent & token → Auditor scores. 16 workers, Stripe, MCP.
- Serica (goserica.com) — Founder. China-inbound travel platform converting global curiosity about China into bookable, AI-generated trips.
Awards & Honors
Public Value Award, Memory Genesis Competition — Evermind Hackathon 2026 · National Scholarship 国家奖学金 (2023, ~0.2% of students) · Gold Award, China Collegiate Programming Contest (CCPC) 2024 · Silver Award, CCPC 2023 · 2nd Prize, China Int'l College Students' Innovation & Entrepreneurship Competition (“Internet+”) · Silver, University Physics Competition 2023 · Honorable Mention, MCM 2023.
AEGIS named among the funded open-source projects in USC Fortis Lab's NVIDIA Academic Grant Program (2026); AEGIS & Sovereign-OS among the named projects in the Amazon Research Award for AI auditing & agent security.
Technical Skills
Languages: Python, C++, TypeScript, Java, CUDA ·
ML/AI: PyTorch, LibTorch, TensorRT, LLM agents, RL, RAG & memory, multi-agent orchestration (LangGraph) ·
Systems: low-latency inference, ZeroMQ, MCP, Stripe, distributed pipelines