Short write-ups on agents, storage, and research.https://iyildirim.com/https://iyildirim.com/notes/tracing-llm-agent-syscalls/https://iyildirim.com/notes/tracing-llm-agent-syscalls/A worked example. I trace a small tool-calling agent on a local model with DFTracer’s agent spans, capture each tool’s system calls in a file of its own, and grow the data 16× twice to see which layer explains what.Sat, 31 Oct 2026 00:00:00 GMThttps://iyildirim.com/notes/capturing-analyzing-fine-grained-dl-io/https://iyildirim.com/notes/capturing-analyzing-fine-grained-dl-io/A worked example. I trace a 3D UNet’s training with DFTracer at four data-loader settings and read the traces with DFAnalyzer to answer the question that matters most: how much of the loading does the GPU actually wait for?Sat, 03 Oct 2026 00:00:00 GMThttps://iyildirim.com/notes/llm-thinking-time-prefetch/https://iyildirim.com/notes/llm-thinking-time-prefetch/An agent spends most of a session waiting on storage, and the seconds it spends reasoning beforehand say which files it is about to read. AgentStage uses that window to stage them.Tue, 29 Sep 2026 00:00:00 GMT