Building trustworthy AI agents, reinforcement learning, and quantitative methods — grounded in rigorous data science.
I'm Peter Fengze Zhong, an AI & Machine Learning researcher at New York University. My work centers on making intelligent agents both capable and trustworthy — pairing modern deep learning with the rigor needed for high-stakes, regulated domains.
Day to day, I work across assured AI agents, reinforcement learning, and quantitative finance, all anchored in careful, reproducible data science. I enjoy taking research ideas from notebook to working system, and care as much about how a model is evaluated and governed as about how it performs.
I build primarily in Python and the Jupyter ecosystem for ML and data science, with TypeScript and JavaScript for interactive applications.
|
🛡️ AI Agents & Assurance Assurance methodology and governance for autonomous AI agents — with an eye toward regulated financial services. |
🤖 Reinforcement Learning Single- and multi-agent RL, including hierarchical and risk-sensitive approaches applied to real-world decision problems. |
|
📈 Quantitative Finance / Trading LLM-infused, risk-aware trading agents and quantitative methods for financial decision-making. |
📊 Data Science End-to-end analysis and modeling — from data wrangling and experimentation to reproducible, well-evaluated pipelines. |
| Project | What it is | Focus / Tech |
|---|---|---|
| agent_assurance | Assurance methodology for autonomous AI agents in regulated financial services. | AI Agents & Assurance · Python |
| Imp_FinRL_DeepSeek | LLM-infused, risk-sensitive reinforcement learning for trading agents. | Quant Finance · RL · Jupyter / Python |
| earth_rover | Image-based autonomous navigation on the FrodoBots Earth Rovers SDK — teleoperation, data logging, and offline analysis. | AI Agents · Robotics · Python |
| citibikeRL | Reinforcement learning on NYC Citi Bike data — RL applied to urban mobility. | Reinforcement Learning · Python |
| nyudsc_pulse | NYU Data Science Club hackathon project. | Data Science · TypeScript / React |
