STUDY OS · ROADMAP v2

The Applied LLM Engineer

Fluency everywhere, plus real depth in the layer where things break.
{{ pct }}%
COMPLETE
{{ t1Done }}/{{ t1Total }}
T1 DEEP
{{ topicsDone }}/{{ topicsTotal }}
ALL TOPICS
{{ projDone }}/{{ projTotal }}
PROJECTS
{{ papersDone }}/{{ papersTotal }}
PAPERS
PHASE {{ col.n }} {{ col.frac }}
{{ col.name }}
{{ col.dur }}
{{ col.note }}
{{ t.tier }}
{{ t.title }}
{{ tag.label }}
{{ t.expandIcon }}
{{ t.idea }}
DONE WHEN
{{ t.exit }}
SOURCE
{{ t.resource }}
▸ BUILD
{{ t.project }}
No topics match — try clearing the search or filters.
MINIMUM VIABLE PROFILE

What you should have shipped

Evidence beats coursework — every item is a real, deployed artifact.
{{ g.label }}
{{ g.sub }}
{{ g.frac }}
{{ it.text }}
WHAT “VERSATILE” ACTUALLY LOOKS LIKE
Not equal skill in every architecture — the judgment to move between them.
{{ v }}
HOW TO RUN THIS

Tiers, exit criteria, retention

A checkbox with no exit criterion is a lie. These three mechanisms are what keep {{ topicsTotal }} topics from becoming {{ topicsTotal }} skimmed blog posts.
{{ m.tag }}
{{ m.title }}
{{ m.body }}
HONEST TIMELINE
{{ r.path }}
{{ r.dur }}
{{ r.scope }}
MUST-READ PAPERS

The canon — all free on arXiv

{{ papersDone }}/{{ papersTotal }} read
3-pass method — 15-min skim, 45-min read, 2–4hr full derivation if it matters to you. After each, close it and write a one-page summary: problem, key insight, implementation, results, limitations.
{{ p.name }}