Slides: Dream Machine: Scaling to 1m users in 4 days — Keegan McCallum, Luma AI
Source Video
Dream Machine: Scaling to 1m users in 4 days — Keegan McCallum, Luma AI
Relationship To World's Fair 2026
These slides are extracted from a public AI Engineer YouTube video connected to World's Fair 2026. Speaker-matched clips are supporting context unless later confirmed as exact session recordings; official livestream recordings are day-level/event-level source material.
Related Scheduled Sessions
- No individual scheduled session mapping has been assigned yet; treat this as an event livestream deck.
Extracted Slides
OCR text:
aWS
ee)
@®Graphite WW Windsurf & MongobB
Mdaily £3augmentcode WorkOS
OCR text:
|
. a
.
mi i 7 5 and PNG J i . B t a 7
; > . } z —_— PS ee “ ;
vad .
a Oe Fe
~ —o
) nN
« Luma
a Microsoft ary?
| —"
OCR text:
amit10:42PM
iamthecha0smonkeyhttps://x.com/LumaLabsAl/status/1801127491496730730
XX(formerlyTwitter)
LumaAl(@LumaLabsAl)onX
Thankyou foryourpatiencewhilewescaled upDreamMachine.It's10xbiggernow！
AIE
Let'sgetbackto imagining...(135kB)
ITIS
RELEASEDAY
MY DUDES.
Luma
Microsoft
smolo
OCR text:
let'sseehowitgoes
KeeganMcCallom 10517
Thomas
KeeganMcCallum 11:31PM
KeeganMcCallm
1952PM
400 ahth ol
AIE
ustPINNING 5000 H100s aIldary
dockerinstalledon1st16/18
pullingimage
Thomus
10S3PM
keepingat aroumd 400muybe
KeeganMeCallm10.5P
Nope oh boy 500
ahhhh this is white knuckling.Ihope these nodes are enough
saanninomscanAi
KeeganMcCallum 12:11AM
YESlol
2500im2vidqueue
banana barely made a dent
Microsoft
smol?
OCR text:
Luma's mission is to build multimodal general
intelligence that can generate, understand, and
operate in the physical world
« Luma
aws
7 |
ae —"
OCR text:
—~_
x
en
B )
c aWws
ed
ee a"
OCR text:
a atl
= =
a ae iomed A a
eon ae Rin at
: re sees 7 : “ee
ce ee) a ”
5 poe a :
a _ Se nae aa
er
= oa a ‘s :
ee
ro
“
|e —
OCR text:
Public AP| ==:
lore
i
= =
https://lumalabs.ai/api/pricing aa
fel a
; a Microsoft ary?
OCR text:
CPU Worker FARR PACE ey CPU Worker a
-
CPU Worker CPU Worker
- ug
triton abe re et ne eee baeeeel COE e See 2 oe
« Luma
—_ 7
si)
a Microsoft §$ou
ee —
OCR text:
Challenges
Brittle,needtocoordinatebetweenbothCPU andTritonbeingupatthesame
AIE
time
Tritonnotbuiltformulti-gpu/multi-node
Pushmodelnotidealformulti-node(whichnodehasrankO?)
No/limitedsupportfornon-nvidiachipsetswithTriton
Verydifficulttodevelopagainst
Needtohaveeverypieceeverywhere,hardtobringindisparatecompute(i.e.
fromourtrainingcluster:kekw:)
Microsoft
smol?
OCR text:
(hapuens)
Redis
API
GpU werkers
Redis
AIE
(stardbg)
Redis
Workers
CPU
Seaweedfs
Luma
aws
OCR text:
Challenges
Backpressure
AIE
Priorities/fairscheduling
Handlingmanydifferentmodels
Handling Bursts
Microsoft
.loms
OCR text:
Queues, Queues, Queues
aw
~ StF. « Luma
. 5 Microsoft Qa?
OCR text:
Model Management
« Luma
a Microsoft ary?
OCR text:
THANK YOU
~_
aaa a « Luma
OCR text:
ON A
v4 a ‘
a ia ———— a oe a a a
tf ,
| LL | |
7 =
a r a
Slide-Derived Subjects To Review
Subject extraction uses video title, related session titles/descriptions, transcript context, and OCR text when available. OCR is best-effort and should be reviewed against the embedded slide images.