GenAIBook/course-pathways.html at main · ApartsinProjects/GenAIBook · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8"/>
<meta content="width=device-width, initial-scale=1" name="viewport"/>
<title>Course Pathways - GenAIBook</title>
<style>
  :root { --bg:#f5efe5; --paper:#fffdfa; --ink:#1d1f1b; --muted:#5d655b; --line:#d9d0c4; --accent:#0f5a48; --accent-2:#8e5622; --soft:#edf5f0; --soft-2:#f6efe4; --shadow:0 18px 54px rgba(29,31,27,.08); --max:1240px; }
  * { box-sizing:border-box; }
  body { margin:0; font-family:Georgia,"Times New Roman",serif; color:var(--ink); line-height:1.62; background:radial-gradient(circle at top right, rgba(15,90,72,.12), transparent 28%), linear-gradient(180deg, #f8f3eb 0%, #f2ecdf 100%); }
  .page { max-width:var(--max); margin:0 auto; padding:28px 20px 64px; }
  .hero, .panel, .path, .closing { background:var(--paper); border:1px solid var(--line); border-radius:24px; box-shadow:var(--shadow); }
  .hero { background:linear-gradient(135deg,#153a32 0%,#1d6955 100%); color:#f8f7f2; padding:40px 36px; margin-bottom:22px; }
  .hero h1 { margin:0 0 10px; font-size:clamp(2.1rem,4vw,3.8rem); line-height:1.05; }
  .hero p { margin:0; max-width:82ch; }
  .topbar { display:flex; flex-wrap:wrap; gap:10px; justify-content:space-between; align-items:center; margin-bottom:18px; }
  .topbar a, .jump, .inline-link { text-decoration:none; color:var(--accent); font-weight:700; }
  .tag { background:var(--soft); color:var(--accent); border-radius:999px; padding:5px 10px; font-size:.84rem; }
  .grid { display:grid; grid-template-columns:320px minmax(0,1fr); gap:22px; align-items:start; }
  .sidebar { position:sticky; top:16px; display:grid; gap:16px; }
  .panel { padding:18px; }
  .panel h2, .closing h2 { margin:0 0 10px; color:var(--accent); text-transform:uppercase; letter-spacing:.08em; font-size:.96rem; }
  .summary { list-style:none; margin:0; padding:0; display:grid; gap:10px; }
  .summary li { background:var(--soft-2); border-radius:14px; padding:10px 12px; font-size:.94rem; }
  .path { padding:24px; margin-bottom:20px; }
  .path h2 { margin:0 0 8px; font-size:1.8rem; }
  .path p { margin:0 0 12px; color:var(--muted); }
  .label { margin:14px 0 6px; color:var(--accent); font-weight:700; text-transform:uppercase; letter-spacing:.06em; font-size:.78rem; }
  ol, ul { margin:0 0 0 20px; padding:0; }
  li { margin-bottom:6px; }
  .week-table { width:100%; border-collapse:collapse; margin-top:8px; font-size:.95rem; }
  .week-table th, .week-table td { border-top:1px solid #e7ddd0; padding:10px 8px; vertical-align:top; text-align:left; }
  .week-table th { color:var(--accent); font-size:.8rem; text-transform:uppercase; letter-spacing:.05em; }
  .closing { padding:24px; }
  .hero-grid { display:grid; grid-template-columns:repeat(3,minmax(0,1fr)); gap:14px; margin-top:18px; }
  .hero-card { background:rgba(255,255,255,.12); border:1px solid rgba(255,255,255,.18); border-radius:18px; padding:14px 16px; }
  .hero-card strong { display:block; margin-bottom:6px; text-transform:uppercase; letter-spacing:.08em; font-size:.8rem; }
  .hero-card span { color:rgba(248,247,242,.93); font-size:.95rem; }
  .editor-note { margin:0 0 16px; padding:16px 18px; border-radius:18px; background:linear-gradient(180deg,#fcfaf5 0%,#f6efe4 100%); border:1px solid #e7dece; }
  .editor-note strong { display:block; margin-bottom:6px; color:var(--accent); text-transform:uppercase; letter-spacing:.06em; font-size:.8rem; }
  @media (max-width:980px) { .hero-grid { grid-template-columns:1fr; } }
  @media (max-width:980px) { .grid { grid-template-columns:1fr; } .sidebar { position:static; } }
</style>
</head>
<body>
<div class="page">
  <header class="hero">
    <div class="tag">Instructor Edition</div>
    <h1>Four 12-Week Course Pathways</h1>
    <p>This page maps the book into four teachable 12-week courses: undergraduate engineering, undergraduate research, graduate engineering, and graduate research. It treats adoption as a design problem, not an afterthought, and shows how the same manuscript can serve different teaching goals without losing coherence.</p>
    <div class="hero-grid">
      <div class="hero-card"><strong>Use</strong><span>Choose a course shape without reverse-engineering the manuscript.</span></div>
      <div class="hero-card"><strong>Audience Fit</strong><span>See how depth, labs, and theory change across undergrad and graduate routes.</span></div>
      <div class="hero-card"><strong>Editorial Payoff</strong><span>Make the book look course-ready to instructors and publishers.</span></div>
    </div>
  </header>

  <div class="topbar">
    <div>
      <a class="inline-link" href="index.html">Back to High-Level TOC</a>
    </div>
    <div>
      <span class="tag">12 Weeks</span>
      <span class="tag">4 Adoption Paths</span>
      <span class="tag">Book Layout Audit</span>
    </div>
  </div>

  <div class="grid">
    <aside class="sidebar">
      <section class="panel">
        <h2>Pathways</h2>
        <ul class="summary">
          <li><a class="inline-link" href="#ug-eng">Undergraduate Engineering</a></li>
          <li><a class="inline-link" href="#ug-res">Undergraduate Research</a></li>
          <li><a class="inline-link" href="#grad-eng">Graduate Engineering</a></li>
          <li><a class="inline-link" href="#grad-res">Graduate Research</a></li>
        </ul>
      </section>
      <section class="panel">
        <h2>Shared Cases</h2>
        <ul class="summary">
          <li><strong>Industrial inspection</strong><br/>best for detection, segmentation, synthetic image generation, and deployment economics.</li>
          <li><strong>Multilingual speech analytics</strong><br/>best for ASR, diarization, synthetic speech augmentation, and streaming inference.</li>
          <li><strong>Multimodal incident review</strong><br/>best for video, audio, retrieval, grounding, structured outputs, and human oversight.</li>
        </ul>
      </section>
      <section class="panel">
        <h2>Adoption Notes</h2>
        <ul class="summary">
          <li>The dependency map now exists; use it to trim or deepen routes without breaking the book arc.</li>
          <li>The main remaining challenge is not coverage but pacing and assignment density.</li>
          <li>Undergraduate routes should point students to the onboarding appendices early.</li>
          <li>Research routes should bring the research-methods appendix and Chapter 08 forward.</li>
          <li><a class="inline-link" href="training-finetuning-toolkit-matrix.html">Use the toolkit matrix</a> when choosing model families and labs for each route.</li>
        </ul>
      </section>
    </aside>

    <main>
      <section class="path" id="ug-eng">
        <h2>Undergraduate Engineering</h2>
        <div class="editor-note"><strong>Design Goal</strong>Keep the course build-heavy, concrete, and paced around manageable system milestones rather than heavy paper volume.</div>
        <p>This version optimizes for system building, manageable cognitive load, and repeatable labs. It should minimize frontier overload and keep research-method complexity secondary to reproducible engineering workflows.</p>
        <div class="label">Who It Serves</div>
        <ul>
          <li>Senior undergraduates in CS, EE, or data/AI programs.</li>
          <li>Students who can code in Python and train models, but are not yet fluent in research reading or formal ablation design.</li>
        </ul>
        <div class="label">12-Week Map</div>
        <table class="week-table">
          <tr><th>Week</th><th>Book Mapping</th><th>Focus</th></tr>
          <tr><td>1</td><td>FM, FM2, 01, 02</td><td>Course orientation, miniature pipeline, choose one recurring case study.</td></tr>
          <tr><td>2</td><td>03, 04, Appendix B/C</td><td>Task framing, toolbox selection, repo setup, smoke tests.</td></tr>
          <tr><td>3</td><td>05, 06</td><td>Data rights, dataset schema, metadata, reproducibility basics.</td></tr>
          <tr><td>4</td><td>07</td><td>Annotation systems, auto-labeling, review queues.</td></tr>
          <tr><td>5</td><td>08, 09</td><td>Baselines, evaluation slices, and the core vision-model track for the chosen project.</td></tr>
          <tr><td>6</td><td>10, 11 or 12</td><td>Audio models plus the relevant generative-model track for the case study.</td></tr>
          <tr><td>7</td><td>15 or 17</td><td>Modality-specialized synthetic data lab: image or audio track.</td></tr>
          <tr><td>8</td><td>16 or 18</td><td>Video or simulation lab depending on course emphasis.</td></tr>
          <tr><td>9</td><td>19, 20</td><td>Judge loops, filtering, stress tests, and hidden evaluation slices.</td></tr>
          <tr><td>10</td><td>21 or 22</td><td>Fine-tuning lab on the chosen modality.</td></tr>
          <tr><td>11</td><td>23, 24</td><td>Build a retrieval-plus-generation system and benchmark deployment options.</td></tr>
          <tr><td>12</td><td>25, 31, 32</td><td>Monitoring, final demo, system card, and capstone presentation.</td></tr>
        </table>
        <div class="label">Book Improvements Needed</div>
        <ul>
          <li>Mark a `core undergraduate engineering path` directly in the front matter.</li>
          <li>Add simpler worked examples for the four-track model section, especially Chapters 09-12.</li>
          <li>Create explicit `minimum viable math` callouts tied to Appendix A instead of leaving students to guess.</li>
        </ul>
      </section>

      <section class="path" id="ug-res">
        <h2>Undergraduate Research</h2>
        <div class="editor-note"><strong>Design Goal</strong>Use the book as a bridge from strong undergraduate implementation skills to careful experimental thinking and paper-style reporting.</div>
        <p>This version is for research seminars, honors projects, or thesis-prep courses. It needs stronger scaffolding around reading papers, forming claims, ablations, and writing defensible reports.</p>
        <div class="label">Who It Serves</div>
        <ul>
          <li>Advanced undergraduates aiming for lab work, a thesis, or early publication-style projects.</li>
          <li>Students who can build systems but need explicit guidance for research design and evaluation logic.</li>
        </ul>
        <div class="label">12-Week Map</div>
        <table class="week-table">
          <tr><th>Week</th><th>Book Mapping</th><th>Focus</th></tr>
          <tr><td>1</td><td>FM, FM2, 01, 03</td><td>Research framing, recurring case studies, project question selection.</td></tr>
          <tr><td>2</td><td>05, 06, 08</td><td>Data rights, dataset design, claim/baseline/ablation planning.</td></tr>
          <tr><td>3</td><td>09, 10</td><td>Vision and audio model internals as preparation for research design.</td></tr>
          <tr><td>4</td><td>11, 12</td><td>Generative models for vision and audio, plus model-selection heuristics for synthetic-data studies.</td></tr>
          <tr><td>5</td><td>07, 13</td><td>Annotation systems plus pre-deployment debugging as research instrumentation.</td></tr>
          <tr><td>6</td><td>14, 20</td><td>Synthetic data engine design and training-vs-evaluation-vs-stress-test distinction.</td></tr>
          <tr><td>7</td><td>15, 16, 17</td><td>Choose one modality branch and build the first study dataset.</td></tr>
          <tr><td>8</td><td>18, 19</td><td>Simulation or judge-loop design depending on the research question.</td></tr>
          <tr><td>9</td><td>21 or 22</td><td>Fine-tuning and measuring real-only vs hybrid vs synthetic-only results.</td></tr>
          <tr><td>10</td><td>23, 24</td><td>System integration and deployment-aware validity concerns.</td></tr>
          <tr><td>11</td><td>25, 29, 30</td><td>Drift, frontier context, and anti-pattern critique.</td></tr>
          <tr><td>12</td><td>31</td><td>Paper-style report, reproducibility package, and presentation.</td></tr>
        </table>
        <div class="label">Book Improvements Needed</div>
        <ul>
          <li>Add a dedicated `research methods primer` appendix or front-matter mini-chapter for paper reading, hypotheses, ablations, and negative results.</li>
          <li>Add explicit literature-note boxes tagged `established`, `best practice`, `emerging`, and `open question`.</li>
          <li>Add a reproducibility checklist and paper-outline template earlier than Chapter 31.</li>
        </ul>
      </section>

      <section class="path" id="grad-eng">
        <h2>Graduate Engineering</h2>
        <div class="editor-note"><strong>Design Goal</strong>Emphasize architecture choice, synthetic-data system design, deployment, and runtime tradeoffs without losing technical rigor.</div>
        <p>This version is for professional MS programs or advanced engineering electives. It should emphasize systems, deployment, cost, and cross-modal architecture decisions, while still being theory-aware.</p>
        <div class="label">Who It Serves</div>
        <ul>
          <li>Graduate students targeting industry ML engineering, applied multimodal AI, or production system design.</li>
          <li>Students comfortable with deep learning internals and able to absorb architecture-heavy chapters faster.</li>
        </ul>
        <div class="label">12-Week Map</div>
        <table class="week-table">
          <tr><th>Week</th><th>Book Mapping</th><th>Focus</th></tr>
          <tr><td>1</td><td>FM, 01, 02, 04</td><td>Book promise, fast-start pipeline, stack selection.</td></tr>
          <tr><td>2</td><td>05, 06, 07</td><td>Data governance, schemas, and annotation operations.</td></tr>
          <tr><td>3</td><td>08, 09</td><td>Experimental rigor and the vision-model track.</td></tr>
          <tr><td>4</td><td>10, 11</td><td>Audio models plus generative models for vision.</td></tr>
          <tr><td>5</td><td>12, 13</td><td>Generative models for audio and debugging instrumentation.</td></tr>
          <tr><td>6</td><td>14, 15</td><td>Synthetic data engine and diffusion-based image pipelines.</td></tr>
          <tr><td>7</td><td>16, 17</td><td>Video and audio synthetic data operations.</td></tr>
          <tr><td>8</td><td>18, 19</td><td>Simulation, QA, reward models, and judge loops.</td></tr>
          <tr><td>9</td><td>20, 21, 22</td><td>Training/eval/stress-test split plus modality fine-tuning decisions.</td></tr>
          <tr><td>10</td><td>23</td><td>Multimodal retrieval, RAG, agent workflows, structured outputs.</td></tr>
          <tr><td>11</td><td>24, 25</td><td>Inference engineering, monitoring, drift, and safe rollback.</td></tr>
          <tr><td>12</td><td>26–28, 31</td><td>Domain blueprint and final system delivery.</td></tr>
        </table>
        <div class="label">Book Improvements Needed</div>
        <ul>
          <li>Add a stronger dependency map so instructors can skip between modalities without breaking coherence.</li>
          <li>Add more explicit decision matrices for model, serving stack, and runtime export choices.</li>
          <li>Add more compact “engineering-only” chapter summaries for programs with limited time for frontier material.</li>
        </ul>
      </section>

      <section class="path" id="grad-res">
        <h2>Graduate Research</h2>
        <div class="editor-note"><strong>Design Goal</strong>Turn the manuscript into a seminar-quality research spine with explicit space for ablations, frontier questions, and reproducibility artifacts.</div>
        <p>This is the most demanding pathway. It should function as a multimodal synthetic-data research seminar plus semester project, with stronger emphasis on theory, evaluation, and open questions.</p>
        <div class="label">Who It Serves</div>
        <ul>
          <li>PhD students, research-track MS students, or advanced seminar cohorts.</li>
          <li>Readers who need not only system recipes but also a rigorous map of methods, assumptions, and unresolved questions.</li>
        </ul>
        <div class="label">12-Week Map</div>
        <table class="week-table">
          <tr><th>Week</th><th>Book Mapping</th><th>Focus</th></tr>
          <tr><td>1</td><td>FM, FM2, 01, 08</td><td>Research positioning, claim discipline, and semester project proposals.</td></tr>
          <tr><td>2</td><td>05, 06, 07</td><td>Data provenance, schema design, annotation/weak-supervision systems.</td></tr>
          <tr><td>3</td><td>09, 10</td><td>Vision-model and audio-model internals.</td></tr>
          <tr><td>4</td><td>11, 12</td><td>Generative models for vision and generative models for audio.</td></tr>
          <tr><td>5</td><td>13</td><td>Pre-deployment debugging and diagnostic instrumentation.</td></tr>
          <tr><td>6</td><td>14, 20</td><td>Synthetic-data engine design and evaluation/stress-test design.</td></tr>
          <tr><td>7</td><td>15–18</td><td>Modality-specific synthetic-data branch with controlled study design.</td></tr>
          <tr><td>8</td><td>19</td><td>Judge loops, calibration-aware filtering, memorization and contamination controls.</td></tr>
          <tr><td>9</td><td>21–23</td><td>Fine-tuning, multimodal systems, and retrieval-grounding comparisons.</td></tr>
          <tr><td>10</td><td>24, 25</td><td>Deployment-aware measurement, drift, and system validity.</td></tr>
          <tr><td>11</td><td>29, 30</td><td>Frontier papers, anti-pattern critique, and negative-results framing.</td></tr>
          <tr><td>12</td><td>31, 32</td><td>Paper-style submission, reproducibility package, and seminar presentation.</td></tr>
        </table>
        <div class="label">Book Improvements Needed</div>
        <ul>
          <li>Add explicit theorem-free but formal research-method boxes where precision matters.</li>
          <li>Add stronger reading lists and benchmark lineage notes per chapter.</li>
          <li>Add a dedicated appendix or companion section for literature review, paper critique, and reproducibility packaging.</li>
        </ul>
      </section>

      <section class="path">
        <h2>Book Layout Improvements Derived From These Four Courses</h2>
        <div class="label">Highest-Value Additions</div>
        <ul>
          <li>Add a front-matter section called `Four 12-Week Adoption Paths` with explicit route maps for the four course types.</li>
          <li>Add chapter-level tags: `Core`, `Advanced`, `Engineering-heavy`, `Research-heavy`, `Undergrad-friendly`, `Graduate-ready`.</li>
          <li>Add a dependency graph so instructors know which chapters can be skipped, compressed, or swapped.</li>
          <li>Add a `Research Methods Companion` section or appendix with paper-reading, ablation, claim-writing, and reproducibility templates.</li>
          <li>Add `minimal prerequisite refresh` callouts at the start of the four-track Part III and at the start of Part IV for students who need math, signal, or transformer refreshers.</li>
          <li>Add `instructor bundles` for each course type: sample pacing, assignments, rubric weight suggestions, and capstone scope guidance.</li>
          <li>Add `course-safe subsets` of the recurring case studies so instructors can choose a smaller implementation track without reauthoring the whole course.</li>
        </ul>
        <div class="label">Most Important Gap</div>
        <p>The book now has enough content to support all four course types, but it still lacks an explicit adoption layer. Without that layer, instructors must reverse-engineer their own path through a rich but ambitious manuscript. That is the main remaining barrier to course adoption.</p>
      </section>

      <section class="closing">
        <h2>Navigation</h2>
        <ul>
          <li><a class="inline-link" href="index.html">Return to the high-level TOC</a></li>
          <li><a class="inline-link" href="front-matter/index.html">Open the front matter</a></li>
        </ul>
      </section>
    </main>
  </div>
</div>
</body>
</html>