#profiling

Performance & Scaling Bug Report

GPU at 30 percent utilisation during training: the data loader was the bottleneck

Ryan Pham · May 6, 2026 · 5 min read

Our image model trained on an expensive GPU that sat idle 70 percent of the time. Profiling showed the GPU waiting on CPU-side JPEG decoding and augmentation in too few DataLoader workers. More workers, pinned memory, GPU-side augmentation and preprocessed shards tripled throughput.

Performance & Scaling

Profile before optimizing: flame graphs for PHP, Node.js and Python in practice

Daniel Tran · Nov 4, 2025 · 7 min read

We once spent two weeks optimizing SQL for an endpoint whose real cost was a logging call. Since then every performance task starts with a profile. Here is how we capture flame graphs in PHP, Node.js and Python, how to read them, and how to measure the win.

Engineering Practice

A repeatable debugging method: reproduce, isolate with git bisect, fix the cause

Daniel Tran · Oct 3, 2024 · 9 min read

Debugging is faster as a method than as inspiration. The six steps we use on every hard bug, from a reliable reproduction to a regression test, walked through on a real slowdown we isolated among 143 commits with an automated git bisect run.