Delyan Kratunov
Member Of Technical Staff at OpenAI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolDelyan Kratunov is a seasoned systems and mobile performance engineer with 16 years of experience, currently a Member of Technical Staff at OpenAI based in San Francisco. He has deep expertise in Android runtime internals and VM stack unwinding—ask him about ART and he'll happily dive into the weeds—alongside a strong track record in Linux kernel tracing (bpf/bpftrace) and low-level C/C++ systems work. At Meta/Facebook he led mobile performance efforts and contributed to production tracing tooling like Facebook's profilo, improving unwinding, synchronization, and filtering for real-world apps. His open-source contributions span libbpf, folly, Redex and Buck, reflecting a focus on cross-platform robustness, build reliability, and lock-free concurrency primitives. Delyan blends pragmatic engineering with research-level curiosity, often tackling subtle platform-specific bugs that make large systems observable and performant. He holds a Computer Science degree from the University of Oxford and thrives on projects where deep systems knowledge materially improves production performance.
16 years of coding experience
10 years of employment as a software developer
Bachelor's Computer Science, Bachelor's Computer Science at University of Oxford
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (45)
code-optimization10
buck10
performance-analytics10
performance-monitor10
c-language10
system-programming10
tracer10
ebpf10
libbpf10
c1110
android-development10
bpf10
java10
performance-measurement10
c1710
Programming languages (9)
Github contributions (5)
facebookarchive/profilo
Nov 2017 - Jan 2022
A library for performance traces from production.
Role in this project:
Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:567 releases, 285 commits, 7 PRs in 4 years 2 months
Contributions summary:Delyan made contributions focused on performance improvements and debugging for the "profilo" library, a production performance tracing tool. Their work included addressing unwinding issues on specific Android versions, fixing synchronization problems, and developing the ability to filter out anonymous mappings. The user's changes spanned the core profiler code, logging components, and build scripts, suggesting involvement in performance optimization across multiple aspects of the library. They introduced the "STRING_NAME" entry type.
performancetracesproduction
bpftrace/bpftrace
Aug 2022 - Nov 2022
High-level tracing language for Linux
Role in this project:Back-end Developer
Back-end DeveloperContributions:37 reviews, 36 commits, 19 PRs in 2 months
Contributions summary:Delyan primarily focused on improving the error handling and output mechanisms of the `bpftrace` tool, ensuring helper errors are correctly piped to the standard output and JSON format. They also addressed issues related to userspace kprobe offset verification, making it more robust and optional. Additionally, the user refactored the code by removing `BpfOrc` and updating the build process to statically link against vendored libraries. These changes enhance the stability and maintainability of the bpftrace codebase.
bpfbccebpftracingtracepoints
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.