Takuya Fukuoka

Education

• 2021/3 Master of Information and Communication Engineering, Graduate School of Information Science and Technology, The University of Tokyo
• 2019/3 Bachelor of Engineering in Information and Communication Engineering, The University of Tokyo

Research

I had belonged to Taura Laboratory for three years from B4 to M2. My research interests are on parallel and distributed computing, high-performance computing, networking (especially Remote Direct Memory Access, RDMA), and System Software.

International Conferences (refereed)

• Takuya Fukuoka, Shigeyuki Sato and Kenjiro Taura: Pitfalls of InfiniBand with On-Demand Paging. International Symposium on Performance Analysis of Systems and Software, ISPASS ‘21, March 2021. acceptance rate: 36.9% (24/65) (Paper) (Slides)

• Takuya Fukuoka, Wataru Endo and Kenjiro Taura: An Efficient Inter-Node Communication System with Lightweight-Thread Scheduling. International Conference on High Performance Computing and Communications, HPCC ‘19, August 2019. acceptance rate: 24.6% (Paper) (Slides)

Dissertations

• Finding and Analyzing Performance Pitfalls of On-Demand Paging of InfiniBand Master Thesis, January 2021. (Thesis) (Slides)
• An Implementation of Inter-node Communication System with Efficient Lightweight-thread Scheduling Bachelor Thesis, February 2019.

Awards

• Best Undergraduate Student Award in 150th SIGOS in Japan, August 2021
• Best Young Presentation Award in xSIG’19, May 2019

Internship Experiences

• Preferred Networks Summer Internship
  • 2019/8 – 2019/9
  • Engaged in a research project related to the optimization of the usage of in-house clusters in PFN.
• Fixstars Long-term Internship
  • 2019/2 – 2020/1
  • Accelerate software using the knowledge of hardware such as SIMD instructions.
• Wantedly Summer Internship (Machine Learning Course)
  • 2017/8
  • Analyzed activity logs with BigQuery and Pandas and found new insight on user’s action which had been unknown.
• Laboro.AI Long-term Internship
  • 2017/5 – 2018/11
  • Developed several enterprise systems with deep learning such as matching engine and similar image search engine.
• Unipro Long-term Internship
  • 2016/12 – 2017/4
  • Developed an internal tool for efficient performance tuning of deep learning framework and developed applications with NLP technologies.

Skills

• C, C++
• Python
• Parallel Programming
  • MPI, OpenMP, Pthread, CUDA, AVX instructions
• RDMA (Remote Direct Memory Access)
  • InfiniBand verbs
• Server Management
  • LDAP, DNS, DHCP, NFS, POP3, SMTP, IPMI

Certifications

• TOEIC 970 (online)
• Network Specialist Examination (IPA, JITEC)
• Database Specialist Examination (IPA, JITEC)
• (Non-registered) Registered Information Security Specialist (IPA, JITEC)

Others

• Blog (in Japanese)
• SlideShare