# Brendan Gregg

> Brendan Gregg is an Australian systems performance engineer at OpenAI, where he works on datacenter optimizations for ChatGPT. He invented flame graphs and the USE Method, pioneered eBPF observability, and is the author of landmark books including 'Systems Performance: Enterprise and the Cloud'. His work is credited with saving the industry over $1 billion in compute costs. Previously an Intel Fellow and performance engineering leader at Netflix, Gregg is one of the most influential engineers in Linux and cloud infrastructure.

- **Role:** Member of Technical Staff (Performance Engineering) at OpenAI
- **Organizations:** OpenAI, Intel (former), Netflix (former), Sun Microsystems (former), Oracle (former), Joyent (former)
- **From:** Newcastle, New South Wales, Australia
- **Nationality:** Australian
- **Education:** Bachelor's degree (Computer Engineering/related), University of Newcastle, Australia
- **Known for:** Invented flame graphs - now the standard visualization for CPU and GPU performance profiling worldwide, Created the USE Method (Utilization, Saturation, Errors) for systematic performance analysis, Pioneered eBPF as an observability technology; authored tracing tools included in multiple operating systems

## Career timeline

- **late 1990s** — Graduated from University of Newcastle, Australia
- **early 2000s** — Early career as independent performance consultant and technical trainer
- **2001** — Photo from server room era, early career in systems
- **mid 2000s** — Joined Sun Microsystems as kernel engineer; developed ZFS L2ARC caching layer
- **2006** — Co-authored 'Solaris Performance and Tools' (Prentice Hall)
- **2011** — Co-authored 'DTrace: Dynamic Tracing in Oracle Solaris, Mac OS X, and FreeBSD' (Prentice Hall)
- **2011** — Joined Joyent (after Oracle acquired Sun)
- **2014** — Joined Netflix as Senior Performance Architect / Leader of Performance Engineering
- **2013** — Received USENIX LISA Outstanding Achievement Award
- **2019** — Published 'BPF Performance Tools: Linux System and Application Observability' (Addison-Wesley)
- **2020** — Published 'Systems Performance: Enterprise and the Cloud, 2nd Edition' (Addison-Wesley)
- **2022** — Left Netflix; joined Intel as Intel Fellow, working on AI/xPU performance
- **2025** — Left Intel
- **2026** — Joined OpenAI as Member of Technical Staff for ChatGPT performance engineering, working remotely from Sydney

## Achievements

- Invented flame graphs - now the standard visualization for CPU and GPU performance profiling worldwide
- Created the USE Method (Utilization, Saturation, Errors) for systematic performance analysis
- Pioneered eBPF as an observability technology; authored tracing tools included in multiple operating systems
- Developed ZFS L2ARC - a pioneering flash memory caching layer for ZFS filesystems
- Received USENIX 2013 LISA Outstanding Achievement Award
- JavaOne Rockstar Speaker Award (2016)
- DockerCon Top Speaker Award (2017)
- Work credited with saving the industry over $1 billion in compute costs
- Authored 4 major technical books on systems performance and tracing
- Delivered keynote addresses at 80+ technical conferences worldwide
- Intel Fellow - the highest technical distinction at Intel
- Holds 2 U.S. patents (assigned to Joyent and Oracle)
- Instructed over 2,000 engineers industry-wide
- Created AI Flame Graphs tool at Intel for GPU performance visualization
- Published flame graph research in Communications of the ACM

## Latest updates

- **2026-02** — Joined OpenAI as Member of Technical Staff working on ChatGPT datacenter performance, based remotely in Sydney, Australia
- **2025-12** — Left Intel after serving as Intel Fellow working on AI and xPU performance
- **2025-11** — Wrote 'Third Stage Engineering' and 'On AI Brendans' blog posts
- **2024-10** — Published AI Flame Graphs tool for GPU performance analysis, initially for Intel Tiber AI Cloud
- **2025-05** — Published Doom GPU Flame Graphs blog post - applying flame graph technique to gaming GPU analysis

## Links

- Website: https://www.brendangregg.com
- LinkedIn: https://www.linkedin.com/in/brendangregg
- Twitter/X: https://x.com/brendangregg
- GitHub: https://github.com/brendangregg

---

Profile page: https://yespress.io/brendan-gregg
Published by YesPress — https://yespress.io
Last updated: 2026-04-22