Gisle Dankel
Principal Software Engineer at Scalemem
Oslo, Norway
Summary
🤩
Rockstar🎓
Top SchoolGisle Dankel is a Principal Software Engineer based in Oslo with eight years of recent experience focused on system and application performance analysis, optimization, and cross-platform virtualization. He has driven observability and profiling initiatives at scale—most notably launching the Kineto project used by the PyTorch profiler and building GPU fleet telemetry and on-demand analysis tooling at Facebook. His background spans dynamic binary translation and VM-level portability from long stints at Transitive, IBM and Intel to shaping IPU observability at Graphcore. Gisle blends low-level systems expertise with practical performance engineering, often tackling memory and runtime issues that unlock substantial efficiency gains. He pairs an M.Eng. in Computer Science with a track record of foundational open-source contributions that underpin modern AI performance tooling.
8 years of coding experience
21 years of employment as a software developer
Mathematics, Mathematics at Volda University College (HVO)M.Eng. Computer Science Computer Languages AI Cpu Design Software Design & Architecture Business, M.Eng. Computer Science Computer Languages AI Cpu Design Software Design & Architecture Business at University of Manchester
Norwegian, Portuguese, German, English
Github Skills (10)
cuda10
performance-analytics10
performance-monitor10
c-language10
performance-monitoring10
performance-analysis10
cprogramming-language10
code-profiling10
profiling10
back-end-development9
Programming languages (5)
Github contributions (5)
pytorch/kineto
Sep 2020 - Nov 2021
A CPU+GPU Profiling library that provides access to timeline traces and hardware performance counters.
Role in this project:
Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:72 reviews, 87 commits, 50 PRs in 1 year 2 months
Contributions summary:Gisle's initial commit introduced the `libkineto` library, laying the foundation for CPU and GPU profiling capabilities. They focused on implementing core functionality, including activity profiling and CUDA runtime activity handling. The contributions also included code changes to address linter errors and fix issues related to memory management, as well as performance optimization.
cudacputracesprofilingtimeline
gdankel/pytorch
Mar 2021 - Nov 2021
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Contributions:16 pushes, 10 branches in 8 months
pythongpu-accelerationdeep-learninggpuacceleration
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.