“Supporting the Brightest Minds in STEM,” it says on the Two Sigma Academic Partnerships website.
No pressure, right?
Not for CS PhD student Yinlin Deng, who has been awarded a Two Sigma PhD Fellowship. She said her research “lies at the intersection of software engineering and machine learning, currently focused on improving various coding tasks with Large Language Models (LLMs). I am interested in developing intelligent analysis, testing, verification, and synthesis techniques to improve the reliability, robustness, and performance of software systems, especially for machine learning systems.
Currently, I focus on automated testing with large language models. For instance, my work, TitanFuzz, demonstrated for the first time that LLMs can be directly applied to fuzz test real-world software systems. To date, my work has led to the discovery of 350+ critical bugs in popular software systems, particularly in popular open-source deep learning libraries including PyTorch, TensorFlow, JAX, and OneFlow.”
Deng is a third-year PhD student in computer science. She received her undergraduate degree from Peking University in 2021.
Deng works with CS professor Lingming Zhang. Her work “aligns closely with Prof. Zhang’s work in software engineering and its interactions with machine learning.” Zhang, impressed by Deng's unique qualities, said, “Since the first time I met Yinlin on Zoom, I could tell she belongs to the very rare type of students who are strong in both theory/mathematics and system engineering.”
Zhang continued,” As a third-year PhD student, she has already published seven research papers in top-tier Software Engineering, Programming Language, and Machine Learning venues. Besides research publications, Yinlin’s work also has made significant practical impacts. For example, she proposed the first-ever approach to directly leveraging Large Language Models (LLMs) for fuzzing real-world systems, a concept that has since emerged as a widely researched topic across Software Engineering, Systems, and Security areas.
I have no doubt that Yinlin has high potential and promise to grow to be a star researcher in the intersection of Software Engineering and Machine Learning (e.g., LLMs for programming/testing).”
Two Sigma Investments is a New York-based hedge fund that applies computational methods, including artificial intelligence, machine learning, and distributed computing, in its trading strategies. Deng noted,
“Although my research doesn’t directly involve processing financial information, in my opinion, it may still be of interest to Two Sigma for several reasons. Firstly, hedge funds like Two Sigma also have a huge software development team, and hence, it is critical to enhance the development workflow to improve software quality and productivity. This is especially true for the financial industry, as any bug or feature delay can incur huge financial losses. Second, hedge funds, including Two Sigma, actively leverage artificial intelligence and machine learning in their trading strategies, and therefore, frontier research in these areas is potentially valuable to them.”
As their also website states, Two Sigma’s Academic Partnerships program aims to support “promising students who are engaged in innovative research in STEM fields. Our mission is to foster inclusive academic communities where academics of all backgrounds feel empowered to expand frontiers in STEM.”
Fellowship recipients will receive an award for two consecutive academic years to cover tuition and fees, plus a stipend for living expenses (totaling $160,000 USD). The award is an unrestricted gift paid directly to the university. They will also receive a one-time award paid directly to them. A Two Sigma Research Mentor will be assigned to each fellowship recipient.