https://lmdlifers.github.io/TonyKoo/research/ Essays on machine learning, agentic systems, cloud architecture and decisions under uncertainty. en https://lmdlifers.github.io/TonyKoo/research/how-coding-agents-find-and-change-code.htmlhttps://lmdlifers.github.io/TonyKoo/research/how-coding-agents-find-and-change-code.htmlA useful coding agent narrows the search, checks its evidence, edits within a boundary and proves the patch worked.Fri, 16 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/from-specification-to-verified-release.htmlhttps://lmdlifers.github.io/TonyKoo/research/from-specification-to-verified-release.htmlA clear specification, reviewed plan and layered tests turn an agent run into an accountable software change.Fri, 16 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/when-an-ml-model-is-ready.htmlhttps://lmdlifers.github.io/TonyKoo/research/when-an-ml-model-is-ready.htmlA strong validation score is a start. Deployment needs a baseline, leakage checks and a plan for changing data.Wed, 14 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/how-data-becomes-reliable-evidence.htmlhttps://lmdlifers.github.io/TonyKoo/research/how-data-becomes-reliable-evidence.htmlA good answer needs a trustworthy record, a search method suited to the question and a path back to the source.Wed, 14 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/how-to-choose-a-system-boundary.htmlhttps://lmdlifers.github.io/TonyKoo/research/how-to-choose-a-system-boundary.htmlDraw the line where state is committed, work is handed off and failures can be retried safely.Wed, 14 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/what-the-cloud-should-own.htmlhttps://lmdlifers.github.io/TonyKoo/research/what-the-cloud-should-own.htmlManaged infrastructure removes some work, but ownership of data, access, recovery and cost still needs a name.Wed, 14 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/how-to-decide-under-uncertainty.htmlhttps://lmdlifers.github.io/TonyKoo/research/how-to-decide-under-uncertainty.htmlA good process records what was known, what was assumed and what would change your mind.Wed, 14 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/how-money-compounds-and-risk-interrupts-it.htmlhttps://lmdlifers.github.io/TonyKoo/research/how-money-compounds-and-risk-interrupts-it.htmlTime helps only when a plan can survive its cash needs, costs and the investor’s own behavior.Wed, 14 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/what-language-models-learn.htmlhttps://lmdlifers.github.io/TonyKoo/research/what-language-models-learn.htmlSeparate learned weights, task context and verification before trying to fix a weak answer.Mon, 12 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/when-agents-need-search.htmlhttps://lmdlifers.github.io/TonyKoo/research/when-agents-need-search.htmlChoose the smallest control loop that improves validated outcomes, and be explicit about what branches.Mon, 12 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/where-autonomy-stops.htmlhttps://lmdlifers.github.io/TonyKoo/research/where-autonomy-stops.htmlLet models propose actions while independent controls decide which effects are allowed.Mon, 12 Oct 2026 16:00:00 GMT https://lmdlifers.github.io/TonyKoo/research/how-to-prove-an-agent-works.htmlhttps://lmdlifers.github.io/TonyKoo/research/how-to-prove-an-agent-works.htmlCompare a strong simple baseline with an agent under the same conditions, then inspect outcomes and costs.Mon, 12 Oct 2026 16:00:00 GMT