logo
HomeScholarship WinnersByteScholars Club
EN中文

2025 Scholarship Winners

CAI Jingwei

CAI Jingwei

Tsinghua University

His research focuses on AI chip architecture and scheduling optimization. As first author (including co-first), four papers have been published at the top four conferences in computer architecture, among which GEMINI received the HPCA 2024 Distinguished Artifact Award (1/410). These results focus on exploring the architectural and scheduling optimization space of high-compute AI Chiplet accelerators. The related open-source tools are widely used in this field and have been commercialized. The research outcomes have significantly advanced research and understanding in this area.

CHEN Huayu

CHEN Huayu

Tsinghua University

His main research area is reinforcement learning, aiming to build a scalable, robust, and adaptive decision engine to free humans from tedious tasks. Currently focusing on unifying reinforcement learning paradigms across different domains (such as vision and language), as well as connecting reinforcement learning with supervised learning theory. First author of NFT and DiffusionNFT series methods.

CHEN Mengzhao

CHEN Mengzhao

The University of Hong Kong

The main research area is efficient inference and training of large models. Participated work has been published in international conferences and journals such as ICLR, NeurIPS, ACL, ICCV, and IJCV. There are six first-author (including co-first) papers. These works cover post-training quantization, quantization-aware training, low-precision training, and co-design of software and hardware, significantly reducing the cost of large model inference and training. The co-design results have been applied in real AI chip businesses.

DAI Sunhao

DAI Sunhao

Renmin University of China

His main research areas are information retrieval and recommender systems, aiming to build responsible and trustworthy intelligent information access systems. More than 20 papers have been published at top international conferences such as SIGIR and KDD, including 12 first/co-first author papers, with over 1,100 Google Scholar citations. A first-author paper was nominated for the SIGIR Best Short Paper Award.
Revealed and systematically explored source bias in information retrieval in the era of foundation models, and organized tutorial sessions on trustworthy information retrieval at conferences such as KDD. Multiple research results (such as OnePiece and RecGPT) have been deployed in real industrial scenarios, significantly improving user experience.

LI Cong

LI Cong

Peking University

Committed to optimizing the performance of data-intensive applications through cross-layer co-design of systems and hardware architectures. As first author (including co-first), published six papers at top conferences such as ISCA, ASPLOS, HPCA, and DAC, and received Best Paper Awards at ISCA and HPCA. These works explore the potential of near-memory computing in general computing and edge-side large model inference, solving key challenges in real-world deployment.

Sharing from Our Scholarship Winners

WEN Chuan
Assistant Professor, Shanghai Jiao Tong UniversityEmbodied AI

"As one of China's most influential corporate scholarships, the ByteDance Scholarship has been a meaningful encouragement in my academic journey. What impressed me most was the review process: the interview felt less like an evaluation and more like an equal, friendly, and inspiring academic exchange. An excellent scholarship selection is not just about filtering achievements, but about understanding and supporting the exploration of young researchers.

Now as a teacher, I especially value its inclusive perspective. The winners' research spans both popular frontier fields and niche directions with long-term potential. This openness encourages young scholars to pursue truly important problems, and ensures that those exploring the frontiers of technology are seen."

DONG Yinpeng
Assistant Professor, Tsinghua UniversityAI Safety and Alignment

"My long-standing research focus is AI safety, a field that defines the boundaries of AI development and supports its broader applications. Receiving the ByteDance Scholarship was a crucial affirmation that strengthened my commitment to this path.

Beyond financial support, the scholarship connects frontier challenges and industry resources with young scholars' growth. In today's rapidly evolving AI landscape, timely trust, resources, and vision can shape how far and deep young researchers go.

To this year's applicants: stay curious, explore the unknown, dare to be different, and pursue the directions you truly care about."

WU Shengqiong
Postdoctoral Researcher, University of OxfordMultimodal Foundation Models and Agents

"After receiving the ByteDance Scholarship, the most unexpected gain was the regular information exchange sessions with frontline industry researchers, where I could discuss cutting-edge AI developments and better understand how academic research translates into practice..

These interactions strengthened my commitment to multimodal understanding and generation, especially improving model stability, controllability, and efficiency in real-world scenarios. I hope my research can deliver practical value and solve real-world problems.

If you also wish to broaden your industry-academia perspective, I highly recommend applying for the ByteDance Scholarship. It is not only an honor, but also a platform connecting outstanding talents with frontier research."

SUN Tianxiang
Founder and CEO of Analemma; Assistant Professor at Shanghai Innovation InstituteAI Research Systems

"Through the ByteDance Scholarship program, I had the opportunity to meet some of the most outstanding young talents in the field of AI. This greatly broadened my horizons and prompted me to think about how to strike a balance between academic ideals and commercial implementation. Today, my research focuses on enabling AI to conduct AI research autonomously, thereby expanding the frontiers of human knowledge across different fields.

The ByteDance Scholarship is widely recognized in both academia and industry for its fairness and high standards. It is not only a "powerful boost" for one's career, but also the starting point of a journey of self-discovery. I hope this year's applicants will have the courage to present the best version of themselves and treat this application as an opportunity to start anew, to think clearly about the questions they truly want to explore, and then pursue them with conviction."

WU Haixu
Postdoctoral Researcher, Massachusetts Institute of Technology (MIT)Deep Learning and Scientific Machine Learning

"When I applied, I was worried that big tech companies might not value AI for Science, but the judges strongly recognized my work. I later realized ByteDance's strong commitment to basic research and even conducted research based on its open-source Protenix.

The ByteDance Scholarship also connected me with peers across fields. Conversations with scholars in LLMs and ML Systems helped me see that AI for Science also needs to embrace 'Scaling up'.

This inspired new research directions, such as domain-specific ML systems and scalable scientific models. Following this path, my collaborators and I recently completed GeoPT, a project that rapidly accumulates physical simulation data through a new pre-training paradigm. I highly encourage students to apply for the ByteDance Scholarship and explore more exciting research directions!"

logo
ByteDance WebsiteByteDance CareerByteDance Seed
Beijing Public Security Network Filing No. 11000002002068 | Beijing ICP Filing No. 12025439-7Privacy Terms
Contact us at ByteDanceScholars@bytedance.com