Slides: From 46% to 90%: Fine-Tuning Tiny LLMs for On-Device Agents — Cormac Brick, Google
Source Video
From 46% to 90%: Fine-Tuning Tiny LLMs for On-Device Agents — Cormac Brick, Google
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:
€3 Braintrust €} WorkOS OpenAl
OCR text:
TLMs:TinyLLMsand
AgentsonEdgeDevices
AIE
Bringing state-of-the-artagenticskilstothe
edgewithopenmodels
Cormac Brick, Principal Engineer, Google Al Edge
GoogleDeepMind
OCR text:
n Agenda 00 Al Edge. SLMs & TLMs
10 Agent Skills locally on Android & iOS
* Sd
oY a a 20 TLM workflow
ey 30 TLMsin Action
Engineering the future of Al
OCR text:
SystemlevelGenAl
AIE
Gemini NanowithAICore
Apple IntelligenceWriting
summarizationonAndroid
toolsonios
Engineering the future of Al
AIEng
OCR text:
NEW
Agent Skills & Gemma 4 E2B &
(‘System GenAl' uses AlCore when avail
Pee 3 .
ae B92 59 eam
. ; Bebe Si.
a a oes Poe S360
nd ~*
OO
Masi ones
Woy pp bore
Oink
Oy 8
P¥AENESA Veacine
WOES cone ve
Tek Can
BRE Gare:
wre
i
. | Al Engineer |
EUROPE
OCR text:
Example Skills - Restaurant Roulette
ere g ;
a
ul : | f
Engineering the future of Al
OCR text:
Example Skills - Restaurant Roulette
Cs Ree Noah aaa
ae
Engineering the future of Al
OCR text:
FunctionGemma TuningLab:Fine-Tuning
atou
tdetalls
Signin with Hugging Face
AIE
2.Tral
Tool Schema&Datalmport
Idas
Step 2:UpldDataIOptin
To tral
onyour
data,up
adaCSv
DropFile Here
Engineering the future of Al
AIEn
OCR text:
TO Sark a |
a -— PC
Sen
i y)
i _ ee He PE ar
q - =e ie
Ey
ach aA
F +n
| ene
1° see
D Lt
nad ;
: ins —
: : “— . ,
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.