Wanqi Zhu

Wanqi Zhu works on coding models at Cursor, including real-time reinforcement learning that uses feedback from production users. He previously worked as a quantitative researcher at Jane Street and writes about learning, software and deciding which problems to pursue.

Career

Cosigned byLoading...

CosignedSign in to view

Based in: San Francisco
Elsewhere

Cursor

Jul 2025 — Present

Research Scientist

Jane Street

Quantitative Researcher
Aug 2021 — Jun 2025
New York, United States
Quantitative Research Intern
Jun 2019 — Aug 2019
New York City, New York, United States

Lyft

Jun 2020 — Aug 2020

Economist Intern

Improbable

Jun 2018 — Sep 2018

Software Engineering Intern

Aura Health

Feb 2018 — Jun 2018

Machine Learning Intern

Stanford University

University of Chicago

Products & media

cursor.com

Improving Composer through Real-Time RL

Coauthored with Jacob Jackson, Ben Trapani and Nathan Wang, this article explains production-feedback training, five-hour checkpoints and preventing reward hacking.

Writing

wanqizhu.substack.com

Directed Acyclic Graph

Wanqi Zhu reflects on learning, career exploration and deciding what progress means when many paths remain open.

Writing

Profile name

Company name

Company name

Profile name

Company name

Company name

Profile name

Company name

Company name

Profile name

Company name

Company name

Profile name

Company name

Company name

Wanqi Zhu

Cursor

Jane Street

Quantitative Researcher

Quantitative Research Intern

Lyft

Improbable

Aura Health

Stanford University

University of Chicago

Products & media

Improving Composer through Real-Time RL

Directed Acyclic Graph

Curated Lists

List Inclusions