-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
1511 lines (1425 loc) · 94.6 KB
/
Copy pathindex.html
File metadata and controls
1511 lines (1425 loc) · 94.6 KB
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
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE html>
<html lang="en">
<head>
<!-- Google tag (gtag.js) -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-62YF7Y81BS"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'G-62YF7Y81BS');
</script>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Omkar Ray</title>
<link rel="stylesheet" href="style.css">
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&family=Newsreader:ital,wght@0,400;0,500;1,400&display=swap" rel="stylesheet">
</head>
<body>
<!-- ─── Daily AI Fact ─── -->
<div class="fact-strip" id="factStrip">
<span class="fact-label">Today's fact</span>
<span class="fact-text" id="factText"></span>
<a class="fact-source" id="factSource" href="#"></a>
<button class="fact-close" id="factClose" type="button" aria-label="Dismiss">×</button>
</div>
<!-- ─── Hero ─── -->
<section class="hero">
<h1 class="hero-name">Omkar Ray</h1>
<p class="hero-bio">Curious, first principles thinker. AMA Music, Tech and Product.</p>
<div class="hero-links">
<a href="https://x.com/OmkarRay9" target="_blank" rel="noopener">X</a>
<a href="https://www.linkedin.com/in/omkarray/" target="_blank" rel="noopener">LinkedIn</a>
<a href="https://omkarray.substack.com/" target="_blank" rel="noopener">Substack</a>
<a href="mailto:omkardray.mml@gmail.com">Email</a>
</div>
</section>
<!-- ─── Spotify Random Song ─── -->
<div class="sp-float" id="spFloat">
<button class="sp-fab" id="spFab" type="button" aria-label="Play a random song">
<svg viewBox="0 0 24 24" fill="currentColor"><path d="M12 0C5.4 0 0 5.4 0 12s5.4 12 12 12 12-5.4 12-12S18.66 0 12 0zm5.521 17.34c-.24.359-.66.48-1.021.24-2.82-1.74-6.36-2.101-10.561-1.141-.418.122-.779-.179-.899-.539-.12-.421.18-.78.54-.9 4.56-1.021 8.52-.6 11.64 1.32.42.18.479.659.301 1.02zm1.44-3.3c-.301.42-.841.6-1.262.3-3.239-1.98-8.159-2.58-11.939-1.38-.479.12-1.02-.12-1.14-.6-.12-.48.12-1.021.6-1.141C9.6 9.9 15 10.561 18.72 12.84c.361.181.54.78.241 1.2zm.12-3.36C15.24 8.4 8.82 8.16 5.16 9.301c-.6.179-1.2-.181-1.38-.721-.18-.601.18-1.2.72-1.381 4.26-1.26 11.28-1.02 15.721 1.621.539.3.719 1.02.419 1.56-.299.421-1.02.599-1.559.3z"/></svg>
</button>
<div class="sp-player" id="spPlayer">
<div class="sp-player-bar">
<button class="sp-shuffle-btn" id="spShuffle" type="button" aria-label="Shuffle to another song">
<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><polyline points="16 3 21 3 21 8"/><line x1="4" y1="20" x2="21" y2="3"/><polyline points="21 16 21 21 16 21"/><line x1="15" y1="15" x2="21" y2="21"/><line x1="4" y1="4" x2="9" y2="9"/></svg>
</button>
<button class="sp-close-btn" id="spClose" type="button" aria-label="Close player">×</button>
</div>
<div class="sp-player-embed" id="spEmbed"></div>
</div>
</div>
<!-- ─── Floppy Section ─── -->
<div class="floppy-section">
<p class="floppy-section-label">What I write about</p>
<!-- Row 1: Deepdives · Thesis · Blogs -->
<div class="floppy-row">
<div class="floppy fade-in" data-folder="deepdives" tabindex="0" role="button" aria-expanded="false">
<div class="floppy-disk">
<svg viewBox="0 0 170 176" fill="none" xmlns="http://www.w3.org/2000/svg">
<rect x="1" y="1" width="168" height="174" rx="8" fill="#E8E5D8" stroke="#D6D3C6" stroke-width="1"/>
<rect class="floppy-shutter" x="36" y="1" width="98" height="42" rx="3" fill="#C8C5B8"/>
<rect x="68" y="7" width="34" height="30" rx="2" fill="#B0ADA0"/>
<rect x="20" y="68" width="130" height="96" rx="4" fill="#fdf0ee" stroke="#e8c4be" stroke-width="0.5"/>
<line x1="30" y1="102" x2="140" y2="102" stroke="#e8c4be" stroke-width="0.4"/>
<line x1="30" y1="116" x2="140" y2="116" stroke="#e8c4be" stroke-width="0.4"/>
<line x1="30" y1="130" x2="140" y2="130" stroke="#e8c4be" stroke-width="0.4"/>
<line x1="30" y1="144" x2="140" y2="144" stroke="#e8c4be" stroke-width="0.4"/>
<text x="85" y="88" text-anchor="middle" font-family="Inter, sans-serif" font-size="13" font-weight="600" fill="#993025">Deepdives</text>
<rect x="4" y="4" width="14" height="14" rx="3" fill="#D6D3C6"/>
</svg>
</div>
<span class="floppy-count">10 explorations</span>
</div>
<div class="floppy fade-in" data-folder="thesis" tabindex="0" role="button" aria-expanded="false">
<div class="floppy-disk">
<svg viewBox="0 0 170 176" fill="none" xmlns="http://www.w3.org/2000/svg">
<rect x="1" y="1" width="168" height="174" rx="8" fill="#E8E5D8" stroke="#D6D3C6" stroke-width="1"/>
<rect class="floppy-shutter" x="36" y="1" width="98" height="42" rx="3" fill="#C8C5B8"/>
<rect x="68" y="7" width="34" height="30" rx="2" fill="#B0ADA0"/>
<rect x="20" y="68" width="130" height="96" rx="4" fill="#eef2f8" stroke="#b8c8de" stroke-width="0.5"/>
<line x1="30" y1="102" x2="140" y2="102" stroke="#b8c8de" stroke-width="0.4"/>
<line x1="30" y1="116" x2="140" y2="116" stroke="#b8c8de" stroke-width="0.4"/>
<line x1="30" y1="130" x2="140" y2="130" stroke="#b8c8de" stroke-width="0.4"/>
<line x1="30" y1="144" x2="140" y2="144" stroke="#b8c8de" stroke-width="0.4"/>
<text x="85" y="88" text-anchor="middle" font-family="Inter, sans-serif" font-size="13" font-weight="600" fill="#1a4a7a">Thesis</text>
<rect x="4" y="4" width="14" height="14" rx="3" fill="#D6D3C6"/>
</svg>
</div>
<span class="floppy-count">3 theses</span>
</div>
<div class="floppy fade-in" data-folder="blogs" tabindex="0" role="button" aria-expanded="false">
<div class="floppy-disk">
<svg viewBox="0 0 170 176" fill="none" xmlns="http://www.w3.org/2000/svg">
<rect x="1" y="1" width="168" height="174" rx="8" fill="#E8E5D8" stroke="#D6D3C6" stroke-width="1"/>
<rect class="floppy-shutter" x="36" y="1" width="98" height="42" rx="3" fill="#C8C5B8"/>
<rect x="68" y="7" width="34" height="30" rx="2" fill="#B0ADA0"/>
<rect x="20" y="68" width="130" height="96" rx="4" fill="#f2f2f0" stroke="#d4d4d0" stroke-width="0.5"/>
<line x1="30" y1="102" x2="140" y2="102" stroke="#d4d4d0" stroke-width="0.4"/>
<line x1="30" y1="116" x2="140" y2="116" stroke="#d4d4d0" stroke-width="0.4"/>
<line x1="30" y1="130" x2="140" y2="130" stroke="#d4d4d0" stroke-width="0.4"/>
<line x1="30" y1="144" x2="140" y2="144" stroke="#d4d4d0" stroke-width="0.4"/>
<text x="85" y="88" text-anchor="middle" font-family="Inter, sans-serif" font-size="13" font-weight="600" fill="#2a2a28">Blogs</text>
<rect x="4" y="4" width="14" height="14" rx="3" fill="#D6D3C6"/>
</svg>
</div>
<span class="floppy-count">5 posts</span>
</div>
</div>
<!-- Folder content: Deepdives -->
<div class="folder-wrapper" data-for="deepdives">
<div class="folder-content">
<div class="folder-header">
<span class="folder-title">Deepdives</span>
<button class="folder-close" aria-label="Close folder">×</button>
</div>
<p class="folder-subtitle">Long-form explorations into topics I find fascinating.</p>
<div class="card-grid">
<a href="rl-as-a-service.html" class="card" data-cat="agents">
<span class="card-tag">RL Infrastructure</span>
<h3>RL-as-a-Service</h3>
<p>Why reinforcement learning infrastructure is the next great platform opportunity — the AWS moment for RL, $7.5B today to $21B+ by 2032.</p>
<span class="card-date">Mar 2026</span>
</a>
<a href="prontofication.html" class="card" data-cat="consumer">
<span class="card-tag">Consumer Tech</span>
<h3>Pronto-fication</h3>
<p>The unbundling of Urban Company — how India's $60B home services market is being decomposed by speed-first vertical startups.</p>
<span class="card-date">2026</span>
</a>
<a href="visa-deepdive.html" class="card" data-cat="fintech">
<span class="card-tag">Fintech</span>
<h3>Visa's Product Ecosystem</h3>
<p>How the world's largest payment network is expanding from cards to a $200T money movement platform.</p>
<span class="card-date">Feb 2026</span>
</a>
<a href="embedded-finance-agents.html" class="card" data-cat="fintech">
<span class="card-tag">Fintech × AI</span>
<h3>Embedded Finance 2.0</h3>
<p>When fintech becomes a function call — the protocol stack being built for agent-first commerce.</p>
<span class="card-date">Feb 2026</span>
</a>
<a href="agent-design-deepdive.html" class="card" data-cat="agents">
<span class="card-tag">AI Agents</span>
<h3>Agent Design Patterns</h3>
<p>Seven context management patterns powering Claude Code, Manus, and Cursor.</p>
<span class="card-date">Feb 2026</span>
</a>
<a href="seeing-like-an-agent.html" class="card" data-cat="agents">
<span class="card-tag">AI Agents</span>
<h3>Seeing Like an Agent</h3>
<p>Tool design, action spaces, and progressive disclosure from Claude Code.</p>
<span class="card-date">Feb 2026</span>
</a>
<a href="india-ecomm-genai.html" class="card" data-cat="ecomm">
<span class="card-tag">India E-Commerce</span>
<h3>Indian E-Comm × GenAI</h3>
<p>How Flipkart, Swiggy, Meesho, and Amazon India deploy GenAI across commerce.</p>
<span class="card-date">Feb 2026</span>
</a>
<a href="datacenter.html" class="card" data-cat="infra">
<span class="card-tag">Infrastructure</span>
<h3>Building a Datacenter</h3>
<p>Six layers forced into simultaneous transition by Nvidia's GPU power density roadmap.</p>
<span class="card-date">Feb 2026</span>
</a>
<a href="bolna-deepdive.html" class="card" data-cat="voice-ai">
<span class="card-tag">Voice AI</span>
<h3>Bolna AI</h3>
<p>Product capabilities, competitive moat, wrapper analysis for India's leading Voice AI platform.</p>
<span class="card-date">Feb 2026</span>
</a>
<a href="smallest-deepdive.html" class="card" data-cat="voice-ai">
<span class="card-tag">Voice AI</span>
<h3>Smallest.ai</h3>
<p>Full-stack model-company analysis: proprietary ASR, LLM, TTS, and moat layers.</p>
<span class="card-date">Feb 2026</span>
</a>
<a href="ringg-deepdive.html" class="card" data-cat="voice-ai">
<span class="card-tag">Voice AI</span>
<h3>Ringg.ai</h3>
<p>Distribution-first no-code voice platform: wrapper risk, GTM moat, and path forward.</p>
<span class="card-date">Feb 2026</span>
</a>
</div>
</div>
</div>
<!-- Folder content: Thesis -->
<div class="folder-wrapper" data-for="thesis">
<div class="folder-content">
<div class="folder-header">
<span class="folder-title">Thesis</span>
<button class="folder-close" aria-label="Close folder">×</button>
</div>
<p class="folder-subtitle">Structured thinking on themes I'm obsessed with.</p>
<div class="thesis-grid">
<a href="mixpanel-voice-ai.html" class="thesis-card">
<span class="thesis-number">01</span>
<div class="thesis-header">
<span class="thesis-theme">Voice AI</span>
<span class="thesis-arrow">→</span>
</div>
<h3>Mixpanel for Voice AI</h3>
<p>Every voice AI platform today ships call logs and basic metrics. None ships the product analytics layer that actually matters: conversation funnels, turn-level drop-off, barge-in rates, intent resolution heatmaps.</p>
<div class="thesis-tags">
<span>Voice Analytics</span>
<span>Conversation Intelligence</span>
<span>Product-Led Growth</span>
</div>
</a>
<a href="rl-environments.html" class="thesis-card">
<span class="thesis-number">02</span>
<div class="thesis-header">
<span class="thesis-theme">AI Infrastructure</span>
<span class="thesis-arrow">→</span>
</div>
<h3>RL Environments Are the New Data Moat</h3>
<p>The AI industry ran on static data for a decade. That era is ending. As RL scaling consumes environment interactions, not tokens, the companies building the gyms where agents train may capture more durable value.</p>
<div class="thesis-tags">
<span>Reinforcement Learning</span>
<span>AI Infrastructure</span>
<span>Agent Training</span>
</div>
</a>
<a href="physical-ai-india.html" class="thesis-card">
<span class="thesis-number">03</span>
<div class="thesis-header">
<span class="thesis-theme">Physical AI</span>
<span class="thesis-arrow">→</span>
</div>
<h3>Physical AI & Robotics: The India Thesis</h3>
<p>India’s $1.98B robotics market sits at the intersection of a manufacturing capex super-cycle, the world’s fifth-largest rare earth reserves, and 20% of global chip design talent.</p>
<div class="thesis-tags">
<span>Robotics</span>
<span>India Supply Chain</span>
<span>Rare Earth</span>
</div>
</a>
</div>
</div>
</div>
<!-- Folder content: Blogs -->
<div class="folder-wrapper" data-for="blogs">
<div class="folder-content">
<div class="folder-header">
<span class="folder-title">Blogs</span>
<button class="folder-close" aria-label="Close folder">×</button>
</div>
<p class="folder-subtitle">Shorter thoughts, observations, and notes from the field.</p>
<div class="blog-list">
<a href="https://omkarray.substack.com/p/taste-is-the-new-bottleneck" class="blog-item" target="_blank" rel="noopener">
<span class="blog-date">Feb 16, 2026</span>
<h3>Taste Is the New Bottleneck</h3>
<span class="blog-arrow">→</span>
</a>
<a href="https://omkarray.substack.com/p/the-world-model-gold-rush-why-most" class="blog-item" target="_blank" rel="noopener">
<span class="blog-date">Feb 9, 2026</span>
<h3>The World Model Gold Rush: Why Most Will Fail (And Who Will Win)</h3>
<span class="blog-arrow">→</span>
</a>
<a href="https://omkarray.substack.com/p/why-just-use-chatgpt-is-a-strategic" class="blog-item" target="_blank" rel="noopener">
<span class="blog-date">Dec 12, 2025</span>
<h3>Why "Just Use ChatGPT" Is a Strategic Failure</h3>
<span class="blog-arrow">→</span>
</a>
<a href="https://omkarray.substack.com/p/indias-glp-1-moment" class="blog-item" target="_blank" rel="noopener">
<span class="blog-date">Dec 3, 2025</span>
<h3>India's GLP-1 Moment</h3>
<span class="blog-arrow">→</span>
</a>
<a href="https://omkarray.substack.com/p/will-meesho-take-the-10-minute-delivery" class="blog-item" target="_blank" rel="noopener">
<span class="blog-date">Nov 28, 2025</span>
<h3>Will Meesho Take the 10-Minute Delivery Route?</h3>
<span class="blog-arrow">→</span>
</a>
</div>
</div>
</div>
<!-- Row 2: Fintech · AI/ML · Maths -->
<div class="floppy-row">
<div class="floppy fade-in" data-folder="fintech" tabindex="0" role="button" aria-expanded="false">
<div class="floppy-disk">
<svg viewBox="0 0 170 176" fill="none" xmlns="http://www.w3.org/2000/svg">
<rect x="1" y="1" width="168" height="174" rx="8" fill="#E8E5D8" stroke="#D6D3C6" stroke-width="1"/>
<rect class="floppy-shutter" x="36" y="1" width="98" height="42" rx="3" fill="#C8C5B8"/>
<rect x="68" y="7" width="34" height="30" rx="2" fill="#B0ADA0"/>
<rect x="20" y="68" width="130" height="96" rx="4" fill="#eef6f0" stroke="#b8d8be" stroke-width="0.5"/>
<line x1="30" y1="102" x2="140" y2="102" stroke="#b8d8be" stroke-width="0.4"/>
<line x1="30" y1="116" x2="140" y2="116" stroke="#b8d8be" stroke-width="0.4"/>
<line x1="30" y1="130" x2="140" y2="130" stroke="#b8d8be" stroke-width="0.4"/>
<line x1="30" y1="144" x2="140" y2="144" stroke="#b8d8be" stroke-width="0.4"/>
<text x="85" y="88" text-anchor="middle" font-family="Inter, sans-serif" font-size="13" font-weight="600" fill="#1e6b3c">Fintech</text>
<rect x="4" y="4" width="14" height="14" rx="3" fill="#D6D3C6"/>
</svg>
</div>
<span class="floppy-count">Themes & reading</span>
</div>
<div class="floppy fade-in" data-folder="aiml" tabindex="0" role="button" aria-expanded="false">
<div class="floppy-disk">
<svg viewBox="0 0 170 176" fill="none" xmlns="http://www.w3.org/2000/svg">
<rect x="1" y="1" width="168" height="174" rx="8" fill="#E8E5D8" stroke="#D6D3C6" stroke-width="1"/>
<rect class="floppy-shutter" x="36" y="1" width="98" height="42" rx="3" fill="#C8C5B8"/>
<rect x="68" y="7" width="34" height="30" rx="2" fill="#B0ADA0"/>
<rect x="20" y="68" width="130" height="96" rx="4" fill="#f4eef8" stroke="#ccb8de" stroke-width="0.5"/>
<line x1="30" y1="102" x2="140" y2="102" stroke="#ccb8de" stroke-width="0.4"/>
<line x1="30" y1="116" x2="140" y2="116" stroke="#ccb8de" stroke-width="0.4"/>
<line x1="30" y1="130" x2="140" y2="130" stroke="#ccb8de" stroke-width="0.4"/>
<line x1="30" y1="144" x2="140" y2="144" stroke="#ccb8de" stroke-width="0.4"/>
<text x="85" y="88" text-anchor="middle" font-family="Inter, sans-serif" font-size="13" font-weight="600" fill="#5a2080">AI / ML</text>
<rect x="4" y="4" width="14" height="14" rx="3" fill="#D6D3C6"/>
</svg>
</div>
<span class="floppy-count">Topics & 80-day plan</span>
</div>
<div class="floppy fade-in" data-folder="maths" tabindex="0" role="button" aria-expanded="false">
<div class="floppy-disk">
<svg viewBox="0 0 170 176" fill="none" xmlns="http://www.w3.org/2000/svg">
<rect x="1" y="1" width="168" height="174" rx="8" fill="#E8E5D8" stroke="#D6D3C6" stroke-width="1"/>
<rect class="floppy-shutter" x="36" y="1" width="98" height="42" rx="3" fill="#C8C5B8"/>
<rect x="68" y="7" width="34" height="30" rx="2" fill="#B0ADA0"/>
<rect x="20" y="68" width="130" height="96" rx="4" fill="#f8f4ec" stroke="#ddd0b0" stroke-width="0.5"/>
<line x1="30" y1="102" x2="140" y2="102" stroke="#ddd0b0" stroke-width="0.4"/>
<line x1="30" y1="116" x2="140" y2="116" stroke="#ddd0b0" stroke-width="0.4"/>
<line x1="30" y1="130" x2="140" y2="130" stroke="#ddd0b0" stroke-width="0.4"/>
<line x1="30" y1="144" x2="140" y2="144" stroke="#ddd0b0" stroke-width="0.4"/>
<text x="85" y="88" text-anchor="middle" font-family="Inter, sans-serif" font-size="13" font-weight="600" fill="#8a5a00">Maths</text>
<rect x="4" y="4" width="14" height="14" rx="3" fill="#D6D3C6"/>
</svg>
</div>
<span class="floppy-count">7 topics</span>
</div>
</div>
<!-- Folder content: Fintech -->
<div class="folder-wrapper" data-for="fintech">
<div class="folder-content">
<div class="folder-header">
<span class="folder-title">Fintech</span>
<button class="folder-close" aria-label="Close folder">×</button>
</div>
<p class="folder-subtitle">Themes I track closely at the intersection of finance, regulation, and AI-native products.</p>
<div class="knowledge-grid">
<div class="knowledge-card">
<div class="knowledge-header">
<span class="knowledge-theme">Areas of Focus</span>
</div>
<div class="knowledge-tags">
<span>Payments Infrastructure</span>
<span>Lending & Underwriting</span>
<span>Risk & Fraud Systems</span>
<span>RegTech & Compliance</span>
<span>Embedded Finance</span>
<span>Agentic Commerce</span>
</div>
<p class="reading-list-label">Reading</p>
<div class="reading-list">
<a href="https://d91labs.substack.com/p/understanding-the-mechanics-of-digital" class="reading-item" target="_blank" rel="noopener">
<span class="reading-title">Understanding the Mechanics of Digital Lending in India</span>
<span class="reading-ext">↗</span>
</a>
<a href="https://d91labs.substack.com/p/decoding-emerging-digital-lending" class="reading-item" target="_blank" rel="noopener">
<span class="reading-title">Decoding Emerging Digital Lending Models in India</span>
<span class="reading-ext">↗</span>
</a>
<a href="https://d91labs.substack.com/p/decoding-the-fintech-revolution-in" class="reading-item" target="_blank" rel="noopener">
<span class="reading-title">Decoding the Fintech Revolution in Underwriting</span>
<span class="reading-ext">↗</span>
</a>
<a href="https://d91labs.substack.com/p/how-fintech-is-solving-for-debt-collections" class="reading-item" target="_blank" rel="noopener">
<span class="reading-title">How Fintech is Solving for Debt Collections</span>
<span class="reading-ext">↗</span>
</a>
<a href="https://www.thepaintedstork.com/p/68-upis-monopoly-problem-infra-ownership" class="reading-item" target="_blank" rel="noopener">
<span class="reading-title">The UPI Bottleneck: Full Stack Ambitions, Zero MDR Reality</span>
<span class="reading-ext">↗</span>
</a>
<a href="https://www.thepaintedstork.com/p/81-credit-lines-on-upi-clou-will" class="reading-item" target="_blank" rel="noopener">
<span class="reading-title">Credit Lines on UPI (CLOU): Will It Eat Small Ticket Lending?</span>
<span class="reading-ext">↗</span>
</a>
</div>
</div>
</div>
</div>
</div>
<!-- Folder content: AI/ML -->
<div class="folder-wrapper" data-for="aiml">
<div class="folder-content">
<div class="folder-header">
<span class="folder-title">AI/ML</span>
<button class="folder-close" aria-label="Close folder">×</button>
</div>
<p class="folder-subtitle">Core machine learning areas I actively study, build with, and write about.</p>
<ul class="topic-list">
<li>
<a href="transformers.html">
<span class="topic-title">Transformers</span>
<span class="topic-subtitle">Attention Is All You Need, and Why</span>
</a>
</li>
<li>
<span>
<span class="topic-title">Reinforcement Learning</span>
<span class="topic-subtitle">Learning from Reward, Not Labels</span>
</span>
</li>
<li>
<span>
<span class="topic-title">AI Agents</span>
<span class="topic-subtitle">Agents That Plan, Act, and Recover</span>
</span>
</li>
<li>
<a href="llm-evals.html">
<span class="topic-title">LLM Evaluation</span>
<span class="topic-subtitle">When Benchmarks Lie: The Maths of LLM Evals</span>
</a>
</li>
<li>
<a href="rag.html">
<span class="topic-title">RAG</span>
<span class="topic-subtitle">Retrieval-Augmented Generation Done Right</span>
</a>
</li>
<li>
<span>
<span class="topic-title">Multimodal AI</span>
<span class="topic-subtitle">One Model, Every Modality</span>
</span>
</li>
<li>
<a href="gradient-descent.html">
<span class="topic-title">Gradient Descent</span>
<span class="topic-subtitle">The Geometry of Gradient Descent</span>
</a>
</li>
<li>
<a href="pytorch.html">
<span class="topic-title">PyTorch</span>
<span class="topic-subtitle">Tensors to Multi-GPU: Visual Map</span>
</a>
</li>
</ul>
<button class="study-plan-toggle" aria-expanded="false">
<span>80 Days LLM & RAG Study Plan</span>
<span class="study-plan-arrow">+</span>
</button>
<div class="study-plan-collapsible">
<div class="phase-label">
<span class="phase-name">Phase 1 · Foundations</span>
<span class="phase-tag">Karpathy micrograd & makemore</span>
</div>
<ul class="topic-list">
<li><a href="llm-day1.html"><span class="topic-title">Day 1</span><span class="topic-subtitle">The Value Class & Computation Graphs</span></a></li>
<li><a href="llm-day2.html"><span class="topic-title">Day 2</span><span class="topic-subtitle">The Backward Pass & Chain Rule</span></a></li>
<li><a href="llm-day3.html"><span class="topic-title">Day 3</span><span class="topic-subtitle">Topological Sort & Full Autograd Engine</span></a></li>
<li><a href="llm-day4.html"><span class="topic-title">Day 4</span><span class="topic-subtitle">Building a Neuron & MLP from Scratch</span></a></li>
<li><a href="llm-day5.html"><span class="topic-title">Day 5</span><span class="topic-subtitle">Training the MLP : Gradient Descent in Action</span></a></li>
<li><a href="llm-day6.html"><span class="topic-title">Day 6</span><span class="topic-subtitle">Intro to Language Modeling : The Bigram Model</span></a></li>
<li><a href="llm-day7.html"><span class="topic-title">Day 7</span><span class="topic-subtitle">Tensors, Broadcasting & torch.Tensor Deep Dive</span></a></li>
<li><a href="llm-day8.html"><span class="topic-title">Day 8</span><span class="topic-subtitle">Training Loops, Loss Functions & Evaluation Splits</span></a></li>
<li><a href="llm-day9.html"><span class="topic-title">Day 9</span><span class="topic-subtitle">MLP Language Model (Bengio et al. 2003)</span></a></li>
<li><a href="llm-day10.html"><span class="topic-title">Day 10</span><span class="topic-subtitle">Embeddings, Learning Rate Schedules & Hyperparameters</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Phase 2 · Deep Networks</span>
<span class="phase-tag">makemore & GPT</span>
</div>
<ul class="topic-list">
<li><a href="llm-day11.html"><span class="topic-title">Day 11</span><span class="topic-subtitle">Activations & Gradients : The Fragility of Deep Nets</span></a></li>
<li><a href="llm-day12.html"><span class="topic-title">Day 12</span><span class="topic-subtitle">Batch Normalization : Taming Internal Covariate Shift</span></a></li>
<li><a href="llm-day13.html"><span class="topic-title">Day 13</span><span class="topic-subtitle">Becoming a Backprop Ninja : Manual Tensor Backprop</span></a></li>
<li><a href="llm-day14.html"><span class="topic-title">Day 14</span><span class="topic-subtitle">Cross-Entropy, Softmax & Classification Gradients</span></a></li>
<li><a href="llm-day15.html"><span class="topic-title">Day 15</span><span class="topic-subtitle">WaveNet Architecture : Dilated Causal Convolutions</span></a></li>
<li><a href="llm-day16.html"><span class="topic-title">Day 16</span><span class="topic-subtitle">Self-Attention from Scratch</span></a></li>
<li><a href="llm-day17.html"><span class="topic-title">Day 17</span><span class="topic-subtitle">Multi-Head Attention & Positional Encoding</span></a></li>
<li><a href="llm-day18.html"><span class="topic-title">Day 18</span><span class="topic-subtitle">The Transformer Block : LayerNorm, Residuals & FFN</span></a></li>
<li><a href="llm-day19.html"><span class="topic-title">Day 19</span><span class="topic-subtitle">Building GPT from Scratch : Full Architecture</span></a></li>
<li><a href="llm-day20.html"><span class="topic-title">Day 20</span><span class="topic-subtitle">Training GPT on Shakespeare : Generation & Sampling</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Phase 3 · LLM Architecture Deep Dive</span>
<span class="phase-tag">Raschka · LLMs From Scratch</span>
</div>
<ul class="topic-list">
<li><a href="llm-day21.html"><span class="topic-title">Day 21</span><span class="topic-subtitle">Tokenization : Byte Pair Encoding from Scratch</span></a></li>
<li><a href="llm-day22.html"><span class="topic-title">Day 22</span><span class="topic-subtitle">Data Preparation & DataLoaders for LLMs</span></a></li>
<li><a href="llm-day23.html"><span class="topic-title">Day 23</span><span class="topic-subtitle">Token Embeddings & Positional Embeddings</span></a></li>
<li><a href="llm-day24.html"><span class="topic-title">Day 24</span><span class="topic-subtitle">Causal Self-Attention : Masked & Scaled</span></a></li>
<li><a href="llm-day25.html"><span class="topic-title">Day 25</span><span class="topic-subtitle">The Full GPT-2 Architecture : Layer by Layer</span></a></li>
<li><a href="llm-day26.html"><span class="topic-title">Day 26</span><span class="topic-subtitle">Pre-training Setup : Weight Init & LR Warmup</span></a></li>
<li><a href="llm-day27.html"><span class="topic-title">Day 27</span><span class="topic-subtitle">Pre-training a Small GPT on Text Data</span></a></li>
<li><a href="llm-day28.html"><span class="topic-title">Day 28</span><span class="topic-subtitle">Generating Text : Temperature, Top-k & Nucleus Sampling</span></a></li>
<li><a href="llm-day29.html"><span class="topic-title">Day 29</span><span class="topic-subtitle">Text Classification with LLMs</span></a></li>
<li><a href="llm-day30.html"><span class="topic-title">Day 30</span><span class="topic-subtitle">Instruction Finetuning : Dataset Format & Preparation</span></a></li>
<li><a href="llm-day31.html"><span class="topic-title">Day 31</span><span class="topic-subtitle">Supervised Finetuning (SFT) in Practice</span></a></li>
<li><a href="llm-day32.html"><span class="topic-title">Day 32</span><span class="topic-subtitle">LoRA : Parameter-Efficient Finetuning</span></a></li>
<li><a href="llm-day33.html"><span class="topic-title">Day 33</span><span class="topic-subtitle">RLHF : Reward Models & PPO Basics</span></a></li>
<li><a href="llm-day34.html"><span class="topic-title">Day 34</span><span class="topic-subtitle">DPO : Direct Preference Optimization</span></a></li>
<li><a href="llm-day35.html"><span class="topic-title">Day 35</span><span class="topic-subtitle">LLM Evaluation : Perplexity, Benchmarks & Human Eval</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Phase 4 · RAG & Retrieval Systems</span>
<span class="phase-tag">LlamaIndex · LangChain · Google NotebookLM</span>
</div>
<ul class="topic-list">
<li><a href="llm-day36.html"><span class="topic-title">Day 36</span><span class="topic-subtitle">Retrieval-Augmented Generation : Why Parametric Memory Isn't Enough</span></a></li>
<li><a href="llm-day37.html"><span class="topic-title">Day 37</span><span class="topic-subtitle">Dense vs. Sparse Retrieval : BM25, TF-IDF & Bi-Encoders</span></a></li>
<li><a href="llm-day38.html"><span class="topic-title">Day 38</span><span class="topic-subtitle">Sentence Transformers & Semantic Embedding Spaces</span></a></li>
<li><a href="llm-day39.html"><span class="topic-title">Day 39</span><span class="topic-subtitle">Vector Databases : Pinecone, Weaviate & Chroma Internals</span></a></li>
<li><a href="llm-day40.html"><span class="topic-title">Day 40</span><span class="topic-subtitle">Chunking Strategies : Fixed, Semantic & Hierarchical</span></a></li>
<li><a href="llm-day41.html"><span class="topic-title">Day 41</span><span class="topic-subtitle">Building a Naive RAG Pipeline End-to-End</span></a></li>
<li><a href="llm-day42.html"><span class="topic-title">Day 42</span><span class="topic-subtitle">Advanced Retrieval : HyDE, MMR & Contextual Compression</span></a></li>
<li><a href="llm-day43.html"><span class="topic-title">Day 43</span><span class="topic-subtitle">Re-ranking with Cross-Encoders : Cohere & FlashRank</span></a></li>
<li><a href="llm-day44.html"><span class="topic-title">Day 44</span><span class="topic-subtitle">LlamaIndex : Document Loaders, Indices & Query Engines</span></a></li>
<li><a href="llm-day45.html"><span class="topic-title">Day 45</span><span class="topic-subtitle">Google NotebookLM : Source-Grounded Reasoning & Audio Overviews</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Phase 5 · Agents & Production Systems</span>
<span class="phase-tag">LangChain · OpenAI · Anthropic</span>
</div>
<ul class="topic-list">
<li><a href="llm-day46.html"><span class="topic-title">Day 46</span><span class="topic-subtitle">Evaluating RAG : RAGAS, Faithfulness & Answer Relevance</span></a></li>
<li><a href="llm-day47.html"><span class="topic-title">Day 47</span><span class="topic-subtitle">Prompt Engineering : Zero-shot, Few-shot & Chain-of-Thought</span></a></li>
<li><a href="llm-day48.html"><span class="topic-title">Day 48</span><span class="topic-subtitle">System Prompts, Personas & Context Window Management</span></a></li>
<li><a href="llm-day49.html"><span class="topic-title">Day 49</span><span class="topic-subtitle">Function Calling & Tool Use : OpenAI, Anthropic & Gemini</span></a></li>
<li><a href="llm-day50.html"><span class="topic-title">Day 50</span><span class="topic-subtitle">LLM Agents : ReAct, Plan-and-Execute & Memory Architectures</span></a></li>
<li><a href="llm-day51.html"><span class="topic-title">Day 51</span><span class="topic-subtitle">Multi-Agent Systems : CrewAI, AutoGen & Coordination Patterns</span></a></li>
<li><a href="llm-day52.html"><span class="topic-title">Day 52</span><span class="topic-subtitle">Structured Outputs : JSON Mode, Instructor & Outlines</span></a></li>
<li><a href="llm-day53.html"><span class="topic-title">Day 53</span><span class="topic-subtitle">Guardrails & Safety : NeMo Guardrails & Llama Guard</span></a></li>
<li><a href="llm-day54.html"><span class="topic-title">Day 54</span><span class="topic-subtitle">LLM Observability : LangSmith, Langfuse & Tracing</span></a></li>
<li><a href="llm-day55.html"><span class="topic-title">Day 55</span><span class="topic-subtitle">Cost & Latency Optimization : Caching, Batching & Model Routing</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Phase 6 · Multimodal & Frontier Models</span>
<span class="phase-tag">GPT-4o · Gemini · Claude 3.5</span>
</div>
<ul class="topic-list">
<li><a href="llm-day56.html"><span class="topic-title">Day 56</span><span class="topic-subtitle">Vision Language Models : LLaVA, GPT-4o & Gemini Vision</span></a></li>
<li><a href="llm-day57.html"><span class="topic-title">Day 57</span><span class="topic-subtitle">Multimodal RAG : Tables, Images & PDFs</span></a></li>
<li><a href="llm-day58.html"><span class="topic-title">Day 58</span><span class="topic-subtitle">Audio & Speech : Whisper, TTS & Real-time Voice</span></a></li>
<li><a href="llm-day59.html"><span class="topic-title">Day 59</span><span class="topic-subtitle">Long Context Models : Gemini 1.5, Claude 200k & Needle-in-Haystack</span></a></li>
<li><a href="llm-day60.html"><span class="topic-title">Day 60</span><span class="topic-subtitle">Mixture of Experts : Mixtral Architecture & Routing Mechanisms</span></a></li>
<li><a href="llm-day61.html"><span class="topic-title">Day 61</span><span class="topic-subtitle">Reasoning Models : Chain-of-Thought, o1 & DeepSeek-R1</span></a></li>
<li><a href="llm-day62.html"><span class="topic-title">Day 62</span><span class="topic-subtitle">Code Generation : DeepSeek Coder, Qwen & GitHub Copilot</span></a></li>
<li><a href="llm-day63.html"><span class="topic-title">Day 63</span><span class="topic-subtitle">Image Generation : Stable Diffusion, DALL-E 3 & Flux</span></a></li>
<li><a href="llm-day64.html"><span class="topic-title">Day 64</span><span class="topic-subtitle">Video & World Models : Sora, Veo & VideoLLMs</span></a></li>
<li><a href="llm-day65.html"><span class="topic-title">Day 65</span><span class="topic-subtitle">Frontier Model Internals : GPT-4, Claude 3.5 & Gemini Ultra</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Phase 7 · Fine-tuning & Alignment</span>
<span class="phase-tag">QLoRA · DPO · Constitutional AI</span>
</div>
<ul class="topic-list">
<li><a href="llm-day66.html"><span class="topic-title">Day 66</span><span class="topic-subtitle">Full Fine-tuning vs. PEFT : When Each Makes Sense</span></a></li>
<li><a href="llm-day67.html"><span class="topic-title">Day 67</span><span class="topic-subtitle">QLoRA : 4-bit Quantized Low-Rank Adaptation</span></a></li>
<li><a href="llm-day68.html"><span class="topic-title">Day 68</span><span class="topic-subtitle">ORPO & SimPO : Next-Generation Alignment Without Reference Models</span></a></li>
<li><a href="llm-day69.html"><span class="topic-title">Day 69</span><span class="topic-subtitle">Constitutional AI & RLAIF : Scaling Alignment with AI Feedback</span></a></li>
<li><a href="llm-day70.html"><span class="topic-title">Day 70</span><span class="topic-subtitle">Synthetic Data Generation : Self-Instruct & Evol-Instruct</span></a></li>
<li><a href="llm-day71.html"><span class="topic-title">Day 71</span><span class="topic-subtitle">Continual Learning & Catastrophic Forgetting in LLMs</span></a></li>
<li><a href="llm-day72.html"><span class="topic-title">Day 72</span><span class="topic-subtitle">Model Merging : SLERP, TIES & DARE</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Phase 8 · Inference, Deployment & LLMOps</span>
<span class="phase-tag">vLLM · Ollama · MLX</span>
</div>
<ul class="topic-list">
<li><a href="llm-day73.html"><span class="topic-title">Day 73</span><span class="topic-subtitle">Quantization : INT8, INT4, GPTQ & AWQ</span></a></li>
<li><a href="llm-day74.html"><span class="topic-title">Day 74</span><span class="topic-subtitle">KV Cache & Speculative Decoding : Speeding Up Autoregression</span></a></li>
<li><a href="llm-day75.html"><span class="topic-title">Day 75</span><span class="topic-subtitle">vLLM & PagedAttention : High-Throughput LLM Serving</span></a></li>
<li><a href="llm-day76.html"><span class="topic-title">Day 76</span><span class="topic-subtitle">Ollama & llama.cpp : Running LLMs Locally on Apple Silicon</span></a></li>
<li><a href="llm-day77.html"><span class="topic-title">Day 77</span><span class="topic-subtitle">Model Compression : Pruning, Distillation & Sparse Attention</span></a></li>
<li><a href="llm-day78.html"><span class="topic-title">Day 78</span><span class="topic-subtitle">Edge Deployment : MLX, ONNX & WebLLM in the Browser</span></a></li>
<li><a href="llm-day79.html"><span class="topic-title">Day 79</span><span class="topic-subtitle">LLMOps : CI/CD for Models, Versioning & Drift Detection</span></a></li>
<li><a href="llm-day80.html"><span class="topic-title">Day 80</span><span class="topic-subtitle">Capstone : End-to-End AI-Native Product Architecture</span></a></li>
</ul>
</div>
<a href="agentic-guide.html" class="ax-banner" style="margin-top:1.5rem;">
<div class="ax-banner-left">
<div>
<div class="ax-banner-title">The Agentic Engineering Guide</div>
<div class="ax-banner-desc">10 parts · 33 chapters on AI agent infrastructure — context engineering, security, MCP, orchestration, observability, and production ops.</div>
</div>
</div>
<span class="ax-banner-arrow">→</span>
</a>
<button class="study-plan-toggle" aria-expanded="false">
<span>The Agentic Engineering Guide — All 33 Chapters</span>
<span class="study-plan-arrow">+</span>
</button>
<div class="study-plan-collapsible">
<div class="phase-label">
<span class="phase-name">Part 1 · Foundations</span>
<span class="phase-tag">3 chapters</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch01"><span class="topic-title">Ch 01</span><span class="topic-subtitle">The Agentic Engineering Landscape</span></a></li>
<li><a href="agentic-guide.html#ch02"><span class="topic-title">Ch 02</span><span class="topic-subtitle">The Capability Jump</span></a></li>
<li><a href="agentic-guide.html#ch03"><span class="topic-title">Ch 03</span><span class="topic-subtitle">The Agentic AI Foundation & Standards</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Part 2 · Context Engineering</span>
<span class="phase-tag">3 chapters</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch04"><span class="topic-title">Ch 04</span><span class="topic-subtitle">Context Windows</span></a></li>
<li><a href="agentic-guide.html#ch05"><span class="topic-title">Ch 05</span><span class="topic-subtitle">The Context Engineering Stack</span></a></li>
<li><a href="agentic-guide.html#ch06"><span class="topic-title">Ch 06</span><span class="topic-subtitle">RAG vs. Agentic Search</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Part 3 · Security & Authorization</span>
<span class="phase-tag">4 chapters</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch07"><span class="topic-title">Ch 07</span><span class="topic-subtitle">The Agent Security Crisis</span></a></li>
<li><a href="agentic-guide.html#ch08"><span class="topic-title">Ch 08</span><span class="topic-subtitle">Zanzibar for AI Agents</span></a></li>
<li><a href="agentic-guide.html#ch09"><span class="topic-title">Ch 09</span><span class="topic-subtitle">Prompt Injection & Tool Poisoning</span></a></li>
<li><a href="agentic-guide.html#ch10"><span class="topic-title">Ch 10</span><span class="topic-subtitle">Sandboxing & Runtime Protection</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Part 4 · Protocols & Standards</span>
<span class="phase-tag">3 chapters</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch11"><span class="topic-title">Ch 11</span><span class="topic-subtitle">Model Context Protocol (MCP)</span></a></li>
<li><a href="agentic-guide.html#ch12"><span class="topic-title">Ch 12</span><span class="topic-subtitle">Agent-to-Agent Communication (A2A)</span></a></li>
<li><a href="agentic-guide.html#ch13"><span class="topic-title">Ch 13</span><span class="topic-subtitle">AGENTS.md & Codebase Onboarding</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Part 5 · Observability & Operations</span>
<span class="phase-tag">3 chapters</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch14"><span class="topic-title">Ch 14</span><span class="topic-subtitle">Agent Traces & OpenTelemetry</span></a></li>
<li><a href="agentic-guide.html#ch15"><span class="topic-title">Ch 15</span><span class="topic-subtitle">Cost Tracking & Token Economics</span></a></li>
<li><a href="agentic-guide.html#ch16"><span class="topic-title">Ch 16</span><span class="topic-subtitle">Incident Response for Agent Failures</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Part 6 · Agent Orchestration</span>
<span class="phase-tag">3 chapters</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch17"><span class="topic-title">Ch 17</span><span class="topic-subtitle">The Agent Loop</span></a></li>
<li><a href="agentic-guide.html#ch18"><span class="topic-title">Ch 18</span><span class="topic-subtitle">Multi-Agent Systems</span></a></li>
<li><a href="agentic-guide.html#ch19"><span class="topic-title">Ch 19</span><span class="topic-subtitle">Agent Memory & Checkpoints</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Part 7 · Team Practices</span>
<span class="phase-tag">3 chapters</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch20"><span class="topic-title">Ch 20</span><span class="topic-subtitle">AI Fatigue & Sustainable Adoption</span></a></li>
<li><a href="agentic-guide.html#ch21"><span class="topic-title">Ch 21</span><span class="topic-subtitle">The Conductor Model</span></a></li>
<li><a href="agentic-guide.html#ch22"><span class="topic-title">Ch 22</span><span class="topic-subtitle">Agent Maturity Model</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Part 8 · Production Workflows</span>
<span class="phase-tag">3 chapters</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch23"><span class="topic-title">Ch 23</span><span class="topic-subtitle">Your First Agent in Production</span></a></li>
<li><a href="agentic-guide.html#ch24"><span class="topic-title">Ch 24</span><span class="topic-subtitle">Security Checklist & Review Policies</span></a></li>
<li><a href="agentic-guide.html#ch25"><span class="topic-title">Ch 25</span><span class="topic-subtitle">Measuring Agent Impact</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Part 9 · Production Engineering</span>
<span class="phase-tag">6 chapters</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch26"><span class="topic-title">Ch 26</span><span class="topic-subtitle">Agent Evaluation & Testing</span></a></li>
<li><a href="agentic-guide.html#ch27"><span class="topic-title">Ch 27</span><span class="topic-subtitle">Enterprise Adoption & Vendor Strategy</span></a></li>
<li><a href="agentic-guide.html#ch28"><span class="topic-title">Ch 28</span><span class="topic-subtitle">Agent Cost Control & FinOps</span></a></li>
<li><a href="agentic-guide.html#ch29"><span class="topic-title">Ch 29</span><span class="topic-subtitle">Agent Governance & Compliance</span></a></li>
<li><a href="agentic-guide.html#ch30"><span class="topic-title">Ch 30</span><span class="topic-subtitle">Structured Outputs & Function Calling</span></a></li>
<li><a href="agentic-guide.html#ch31"><span class="topic-title">Ch 31</span><span class="topic-subtitle">Model Selection & Routing</span></a></li>
</ul>
<div class="phase-label">
<span class="phase-name">Part 10 · Sustainability & Deployment</span>
<span class="phase-tag">2 chapters + appendices</span>
</div>
<ul class="topic-list">
<li><a href="agentic-guide.html#ch32"><span class="topic-title">Ch 32</span><span class="topic-subtitle">Backpressure & Automated Feedback</span></a></li>
<li><a href="agentic-guide.html#ch33"><span class="topic-title">Ch 33</span><span class="topic-subtitle">The Agent Adoption Playbook</span></a></li>
<li><a href="https://agents.siddhantkhare.com/appendix-a-tools/" target="_blank" rel="noopener"><span class="topic-title">Appendix A</span><span class="topic-subtitle">Tool & Framework Directory ↗</span></a></li>
<li><a href="https://agents.siddhantkhare.com/appendix-b-glossary/" target="_blank" rel="noopener"><span class="topic-title">Appendix B</span><span class="topic-subtitle">Glossary ↗</span></a></li>
<li><a href="https://agents.siddhantkhare.com/appendix-c-reading/" target="_blank" rel="noopener"><span class="topic-title">Appendix C</span><span class="topic-subtitle">Further Reading & References ↗</span></a></li>
<li><a href="https://agents.siddhantkhare.com/appendix-d-templates/" target="_blank" rel="noopener"><span class="topic-title">Appendix D</span><span class="topic-subtitle">Templates & Checklists ↗</span></a></li>
<li><a href="https://agents.siddhantkhare.com/epilogue/" target="_blank" rel="noopener"><span class="topic-title">Epilogue</span><span class="topic-subtitle">The Next Twelve Months ↗</span></a></li>
</ul>
</div>
</div>
</div>
<!-- Folder content: Maths -->
<div class="folder-wrapper" data-for="maths">
<div class="folder-content">
<div class="folder-header">
<span class="folder-title">Maths</span>
<button class="folder-close" aria-label="Close folder">×</button>
</div>
<p class="folder-subtitle">Mathematical foundations that shape how I reason about models and systems.</p>
<ul class="topic-list">
<li>
<a href="bayesian-optimization.html">
<span class="topic-title">Bayesian Optimization</span>
<span class="topic-subtitle">Probabilistic search & surrogate models</span>
</a>
</li>
<li>
<a href="https://www.youtube.com/playlist?list=PLhoHEZlJjdQKI1cs5yPRUYdgcsE0HctoQ" target="_blank" rel="noopener">
<span class="topic-title">Gaussian Processes</span>
<span class="topic-subtitle">Distributions over functions</span>
</a>
</li>
<li>
<a href="https://www.youtube.com/playlist?list=PLJHszsWbB6hrkmmq57lX8BV-o-YIOFsiG" target="_blank" rel="noopener">
<span class="topic-title">Tensors</span>
<span class="topic-subtitle">Multilinear algebra & geometry</span>
</a>
</li>
<li>
<a href="https://www.youtube.com/playlist?list=PLBh2i93oe2qsGKDOsuVVw-OCAfprrnGfr" target="_blank" rel="noopener">
<span class="topic-title">Hilbert Spaces</span>
<span class="topic-subtitle">Infinite-dimensional inner product spaces</span>
</a>
</li>
<li>
<a href="https://www.youtube.com/playlist?list=PLBh2i93oe2qvRGAtgkTszX7szZDVd6jh1" target="_blank" rel="noopener">
<span class="topic-title">Manifolds</span>
<span class="topic-subtitle">Geometry beyond flat space</span>
</a>
</li>
<li>
<a href="https://www.youtube.com/playlist?list=PLBh2i93oe2qswFOC98oSFc37-0f4S3D4z" target="_blank" rel="noopener">
<span class="topic-title">Probability Theory</span>
<span class="topic-subtitle">Measure-theoretic foundations</span>
</a>
</li>
<li>
<a href="https://www.youtube.com/playlist?list=PLBh2i93oe2quLc5zaxD0WHzQTGrXMwAI6" target="_blank" rel="noopener">
<span class="topic-title">Linear Algebra</span>
<span class="topic-subtitle">Vectors, matrices & transformations</span>
</a>
</li>
</ul>
</div>
</div>
<!-- Row 3: Design -->
<div class="floppy-row">
<div class="floppy fade-in" data-folder="design" tabindex="0" role="button" aria-expanded="false">
<div class="floppy-disk">
<svg viewBox="0 0 170 176" fill="none" xmlns="http://www.w3.org/2000/svg">
<rect x="1" y="1" width="168" height="174" rx="8" fill="#E8E5D8" stroke="#D6D3C6" stroke-width="1"/>
<rect class="floppy-shutter" x="36" y="1" width="98" height="42" rx="3" fill="#C8C5B8"/>
<rect x="68" y="7" width="34" height="30" rx="2" fill="#B0ADA0"/>
<rect x="20" y="68" width="130" height="96" rx="4" fill="#f3eef8" stroke="#c8b8e0" stroke-width="0.5"/>
<line x1="30" y1="102" x2="140" y2="102" stroke="#c8b8e0" stroke-width="0.4"/>
<line x1="30" y1="116" x2="140" y2="116" stroke="#c8b8e0" stroke-width="0.4"/>
<line x1="30" y1="130" x2="140" y2="130" stroke="#c8b8e0" stroke-width="0.4"/>
<line x1="30" y1="144" x2="140" y2="144" stroke="#c8b8e0" stroke-width="0.4"/>
<text x="85" y="88" text-anchor="middle" font-family="Inter, sans-serif" font-size="13" font-weight="600" fill="#6b3fa0">Design</text>
<rect x="4" y="4" width="14" height="14" rx="3" fill="#D6D3C6"/>
</svg>
</div>
<span class="floppy-count">Artifacts</span>
</div>
</div>
<!-- Folder content: Design -->
<div class="folder-wrapper" data-for="design">
<div class="folder-content">
<div class="folder-header">
<span class="folder-title">Design</span>
<button class="folder-close" aria-label="Close folder">×</button>
</div>
<p class="folder-subtitle">Visual artifacts — packaging concepts, interface explorations, and design systems.</p>
<div class="card-grid">
<a href="acai-theory.html" class="card" data-cat="design">
<span class="card-tag">Packaging</span>
<h3>Acai Theory</h3>
<p>A three-panel packaging label for a cold-pressed acai blend — WebGL paper texture, ink-style typography, and risograph-inspired color.</p>
<span class="card-date">Mar 2026</span>
</a>
</div>
<div class="nomad-showcase">
<div class="nomad-showcase-header">
<span class="nomad-showcase-title">Nomad</span>
<span class="nomad-showcase-meta">Design System · 5 Screens · Mar 2026</span>
</div>
<p class="nomad-showcase-desc">A curated travel UI — editorial typography meets pastel cartography. Onboarding through empty states, designed for mobile-first exploration.</p>
<div class="nomad-phone-row">
<a href="/nomad-setup.html" class="nomad-phone-item">
<div class="nomad-phone-frame">
<div class="nomad-phone-notch"></div>
<iframe src="/nomad-setup.html" loading="lazy" scrolling="no"></iframe>
</div>
<span class="nomad-phone-label">Setup</span>
<span class="nomad-phone-sublabel">Travel Identity</span>
</a>
<a href="/nomad.html" class="nomad-phone-item">
<div class="nomad-phone-frame">
<div class="nomad-phone-notch"></div>
<iframe src="/nomad.html" loading="lazy" scrolling="no"></iframe>
</div>
<span class="nomad-phone-label">Journeys</span>
<span class="nomad-phone-sublabel">Curated Feed</span>
</a>
<a href="/nomad-detail.html" class="nomad-phone-item">
<div class="nomad-phone-frame">
<div class="nomad-phone-notch"></div>
<iframe src="/nomad-detail.html" loading="lazy" scrolling="no"></iframe>
</div>
<span class="nomad-phone-label">Umbrian Stone</span>
<span class="nomad-phone-sublabel">Journey Detail</span>
</a>
<a href="/nomad-wishlist.html" class="nomad-phone-item">
<div class="nomad-phone-frame">
<div class="nomad-phone-notch"></div>
<iframe src="/nomad-wishlist.html" loading="lazy" scrolling="no"></iframe>
</div>
<span class="nomad-phone-label">Wishlist</span>
<span class="nomad-phone-sublabel">Saved Journeys</span>
</a>
<a href="/nomad-empty.html" class="nomad-phone-item">
<div class="nomad-phone-frame">
<div class="nomad-phone-notch"></div>
<iframe src="/nomad-empty.html" loading="lazy" scrolling="no"></iframe>
</div>
<span class="nomad-phone-label">No Results</span>
<span class="nomad-phone-sublabel">Empty State</span>
</a>
</div>
</div>
</div>
</div>
</div>
<!-- ─── Must Visits Floating Widget ─── -->
<div class="mv-float" id="mvFloat">
<button class="mv-fab" id="mvFab" type="button" aria-label="Must visits">
<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round"><path d="M2 3h6a4 4 0 0 1 4 4v14a3 3 0 0 0-3-3H2z"/><path d="M22 3h-6a4 4 0 0 0-4 4v14a3 3 0 0 1 3-3h7z"/></svg>
</button>
<div class="mv-panel" id="mvPanel">
<div class="mv-panel-header">
<span class="mv-panel-title">Must visits</span>
<button class="mv-panel-close" id="mvClose" type="button" aria-label="Close">×</button>
</div>
<div class="mv-panel-list">
<a href="https://bestdesignsonx.com/" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">Best Designs On X</span>
<span class="mv-desc">Curated best design work on X.com</span>
</a>
<a href="https://www.blrbloom.com" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">Bengaluru Bloom</span>
<span class="mv-desc">Tabebuia bloom tracker for Bengaluru</span>
</a>
<a href="https://nonint.com/" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">Non_Interactive</span>
<span class="mv-desc">Software & ML — GPT-4o, DALL-E 3, compute multipliers</span>
</a>
<a href="https://jvns.ca/" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">Julia Evans</span>
<span class="mv-desc">Zines & deep explanations on Linux, DNS, Git, networking</span>
</a>
<a href="https://visualpde.com/" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">VisualPDE</span>
<span class="mv-desc">Interactive PDE simulations — waves, diffusion, patterns</span>
</a>
<a href="https://www.feynmanlectures.caltech.edu/I_toc.html" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">Feynman Lectures</span>
<span class="mv-desc">The legendary Caltech physics lectures, free online</span>
</a>
<a href="https://www.isfixable.com/" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">Fixable</span>
<span class="mv-desc">Repairable, open-hardware products for India</span>
</a>
<a href="https://grahamduncan.blog/" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">Graham Duncan</span>
<span class="mv-desc">On talent, infinite games, and letters to friends</span>
</a>
<a href="https://pudding.cool/" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">The Pudding</span>
<span class="mv-desc">Visual essays that explain ideas with data</span>
</a>
<a href="https://blog.ncase.me/" class="mv-item" target="_blank" rel="noopener">
<span class="mv-name">Nicky Case</span>
<span class="mv-desc">Interactive explorable explanations & public-domain art</span>
</a>
</div>
</div>
</div>
<button class="back-to-top" id="backToTop" aria-label="Back to top">↑</button>
<!-- ─── AI Chat Widget ─── -->
<button class="chat-fab" id="chatFab" type="button" aria-label="Ask about Omkar's work">
<svg class="chat-fab-icon" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round"><path d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2z"/></svg>
<span class="chat-fab-ping"></span>
</button>
<div class="chat-window" id="chatWindow">
<div class="chat-header">
<div class="chat-header-left">
<div class="chat-avatar">O<span class="chat-online"></span></div>
<div>
<div class="chat-header-name">Ask Omkar's site</div>
<div class="chat-header-status" id="chatStatus">Online — knows all 10 deep dives</div>
</div>
</div>
<button class="chat-header-close" id="chatClose" type="button" aria-label="Close chat">×</button>
</div>
<div class="chat-body" id="chatBody">
<div class="chat-msg bot chat-animate">
<div class="chat-bubble">Hey! I know everything Omkar has published. Ask me about any topic or try one of these:</div>
</div>
<div class="chat-chips" id="chatChips">
<button type="button" class="chat-chip" data-q="What has Omkar written about voice AI?">Voice AI</button>
<button type="button" class="chat-chip" data-q="Tell me about AI agents">AI Agents</button>
<button type="button" class="chat-chip" data-q="What fintech content is there?">Fintech</button>
<button type="button" class="chat-chip" data-q="Who is Omkar?">About Omkar</button>
<button type="button" class="chat-chip" data-q="Show me everything">All topics</button>
</div>
</div>
<form class="chat-input-row" id="chatForm" autocomplete="off">
<input class="chat-input" id="chatInput" type="text" placeholder="Ask about voice AI, fintech, agents..." aria-label="Your question">
<button class="chat-send" type="submit" aria-label="Send">
<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><line x1="22" y1="2" x2="11" y2="13"/><polygon points="22 2 15 22 11 13 2 9 22 2"/></svg>
</button>
</form>
</div>
<footer>
<p>© 2026 Omkar Ray. Built with care.</p>
</footer>
<script>
(function() {
var facts = [
{ text: "Agent task length doubles every 7 months. Meta bought Manus for $2B+ and Claude Code hit a $1B run rate.", src: "Agent Design Patterns", href: "agent-design-deepdive.html" },
{ text: "Only ~15% of an agent's context window is the actual task. Tool definitions, chat history, and results eat the rest.", src: "Agent Design Patterns", href: "agent-design-deepdive.html" },
{ text: "\"Context engineering is the delicate art of filling the context window with just the right information for the next step.\" — Andrej Karpathy", src: "Agent Design Patterns", href: "agent-design-deepdive.html" },
{ text: "Programmatic Tool Calling gives +11% accuracy and -24% input tokens. Opus 4.6 with PTC is #1 on LMArena Search Arena.", src: "Seeing Like an Agent", href: "seeing-like-an-agent.html" },
{ text: "Claude Code has ~20 tools. The bar to add tool #21 is high — before adding it, ask: can progressive disclosure handle this?", src: "Seeing Like an Agent", href: "seeing-like-an-agent.html" },
{ text: "ChatGPT has 65M daily active users in India — 2x the United States. GenAI app downloads in India: 198M (2024) → 602M (2025).", src: "Indian E-Comm × GenAI", href: "india-ecomm-genai.html" },
{ text: "Meesho's ElevenLabs voice bot handles 60K calls/day with a 95% resolution rate — India's first Gen-AI voice bot at scale.", src: "Indian E-Comm × GenAI", href: "india-ecomm-genai.html" },
{ text: "BharatML: 66.9 trillion feature retrievals in FY25. Meesho open-sourced their production ML platform doing 3.12T inferences at peak.", src: "Indian E-Comm × GenAI", href: "india-ecomm-genai.html" },
{ text: "Alipay AI Pay processes 120M transactions per week. Money movement is often just a row change in their own database.", src: "Embedded Finance 2.0", href: "embedded-finance-agents.html" },
{ text: "\"When a company claims they're building AI-first, I don't ask about their LLM choice. I ask: show me your database schema.\" — Ambika Pande", src: "Embedded Finance 2.0", href: "embedded-finance-agents.html" },
{ text: "Visa's total addressable flows: $200T. Consumer payments are ~$20T. B2B alone is $145T.", src: "Visa Product Ecosystem", href: "visa-deepdive.html" },
{ text: "Visa Direct processed 10B transactions in FY24. 11.5B network tokens provisioned across 60+ countries.", src: "Visa Product Ecosystem", href: "visa-deepdive.html" },
{ text: "Datacenter rack power: 50x increase in 10 years. Volta 20kW/rack (2017) → Rubin Ultra 1,000kW/rack (2027).", src: "Building a Datacenter", href: "datacenter.html" },
{ text: "China controls ~80% of global gallium supply — needed for GaN transistors in datacenter power systems. Export restrictions active since 2025.", src: "Building a Datacenter", href: "datacenter.html" },