CV

EDUCATION BACKGROUND

ETH Zurich 2023.9-present

Department of Computer Science

Direct Doctorate

Expected Date of MSc Graduate: 2025.9

PEKING UNIVERSITY 2019.9-2023.7

School of Electronics Engineering and Computer Science

B.S. in Computer Science

EXPERIENCE

• ASC20-21 Student Supercomputer Challenge 2021.5 Shenzhen, China

  Participate in the competition as a representative of Peking University and win First Prize (team best)

  • Contribution: responsible for PRESTO Challenge - 29-times speedup of a CPU-intensive scientific pulsar-searching toolkit using MPI, OpenMP, Ramdisk and Math kernel library (MKL) SIMD strategies
  • Skills: 1) team training of CUDA, MPI, and VTune, etc. in the broad field of parallel computing and program profiling; 2) specific optimization methodology towards CPU and IO-intensive application; 3) on-site heterogeneous cluster configuration and power control
  • Teamwork: cooperation of profiling and optimization; on-the-spot adaptability of ultra-large datasets

• Internship at DP Technology as Algorithm Researcher 2021.10-Present Beijing, China

  Design and develop Uni-Dock, a GPU-accelerated molecular docking software, integrated in advanced industrial computer-aided drug design (CADD) product

  • Contribution: design and implementation of molecular docking software achieving more than 1600 times acceleration on GPU compared to CPU, enabling billion-scale virtual screening
  • Skills: C++/CUDA programming and tuning techniques, especially in Monte Carlo searches; proficient in profiling hotspots, memory bounds and IO bounds associated with hardware features
  • Usability: engineering algorithmic projects to production, mastering tools such as docker, CI/CD, and unit-test Uni-Dock + ML: combine physical force fields of Uni-Dock with pre-trained molecular representation ML framework, improving molecular docking pose accuracy from 69% to 86%
  • Contribution: 1) design correction of physical force fields using ML to enhance the accuracy of Monte Carlo searches of poses, patented as an invention; 2) improve the probability of hit compound identification in early virtual screening

EXTRACURRICULAR ACTIVITY

• ADA Innovation Fair 2020 2020.8 Online, Toronto

  Develop an application prototype called E-PEDIA for college students course selection during the epidemic

• Contemporary Undergraduate Mathematical Contest in Modeling 2020 2020.9 Online, China

  Win the Second Prize of Beijing division

• The 3rd Priority Research Application Competition 2022.8-2022.10 Beijing, China

  Win the Third Prize in AI for Science track as team leader: migration of Uni-Dock from NVIDIA Volta to ROCm architecture hardware with linear scalability on 1000+ nodes

PUBLICATIONS

• Yu, Y., Cai, C., Wang, J., Bo, Z., Zhu, Z., & Zheng, H. (2023). Uni-Dock: GPU-accelerated docking enables ultralarge virtual screening. Journal of chemical theory and computation, 19(11), 3336-3345.
• Yu, Y., Lu, S., Gao, Z., Zheng, H., & Ke, G. (2023). Do deep learning models really outperform traditional approaches in molecular docking?. arXiv preprint arXiv:2302.07134.
• Huang, L., Gianinazzi, L., Yu, Y., Dueben, P. D., & Hoefler, T. (2024). Diffda: a diffusion model for weather-scale data assimilation. arXiv preprint arXiv:2401.05932.

AWARDS AND CERTIFICATES

• Computer software copyright of Uni-Dock, National Copyright Administration 2022.5 Beijing,China
• Second Class Scholarship, Peking University 2021.9 Beijing,China
• Ubiquant Investment Scholarship, Peking University 2022.9 Beijing,China
• Learning Excellence Award, Peking University 2021.9 Beijing,China
• Outstanding Scientific Research Award, Peking University 2022.9 Beijing,China

SKILLS AND HOBBIES

• Languages——Mandarin (Native), English Fluent (TOEFL 108 S25, GRE 328+4.0)
• Programming Languages——C/C++, CUDA, Python, Bash, JavaScript, HTML
• High Performance Computing——Performance profiling & optimizations, MPI, OpenMP, OpenACC