Learn the frontier · Essays
Thoughtful arguments about where this is all going.
Forward-looking, argument-driven essays. Each leads with a clear thesis and a bet you could actually check — generous thinking-out-loud about where the frontier moves next, written to sharpen how you see your own road ahead, not to chase a trend.
Across 155 small-model experiments centered on Qwen 3.5 4B, the same thing kept working: give the model something executable it can check against the evidence it has, and it punches far above its benchmark weight. Here is the field guide — the levers that worked, how I know they're real, and the frontier they opened up.
Swapping the frontier model rarely moves your agent's success rate as much as fixing retries and context management — and the one thing competitors can't clone is your evaluation environment. A thesis on why agent evals, not weights, are where reproducible capability accrues.
Reasoning models turned inference into a per-request dial. This is an economic read on when spending FLOPs at test time actually buys accuracy, why it only pays where answers are cheap to verify, and what variable-cost inference does to latency budgets and capacity planning.
A practical, friendly guide to reading ML results: why the experimental setup behind a benchmark number matters, how data contamination fools everyone, how to pin your eval harness so scores hold still, and why we build and run things ourselves — so you can learn from them and trust them.