LUMI AI Fac o y Se ice Cen e
Empowe ing Eu ope’s AI Ecosys em
D6.3 AI Fac o y Hub Handbook
2
D6.3
AI Fac o y Hub Handbook
D6.3 AI Fac o y Hub Handbook
3
P ojec Ti le
LUMI AI Fac o y Se ice Cen e
P ojec Ac onym
LUMI AIF
P ojec Numbe
101234208
Type o Ac ion
HORIZON-JU-RIA
Topic
HORIZON-JU-EUROHPC-2025-AI-01-IBA-01
S a ing Da e o P ojec
01.03.2025
Ending Da e o P ojec
29.02.2028
Du a ion o he P ojec
36 mon hs
Websi e
lumi-ai- ac o y.eu
Wo k Package
WP6
Task
T6.5
Lead Au ho s
Ee a Ha jula (CSC), Ka eřina Š id nocho á ( ia IT4I)
Con ibu o s
Ma in Duda (IT4I), Abdul ahman Azab (Sigma2), Ka ja Mankinen
(CSC)
Pee Re iewe s
Alan Medla (UH), Sa a Ga a elli (CSC)
Ve sion
1.0
Due Da e
30.09.2025
Submission Da e
25.09.2025
Dissemina ion le el
x
PU: Public
SEN: Sensi i e – limi ed unde he condi ions o he G an Ag eemen
EU-RES. Classi ied In o ma ion: RESTREINT UE (Commission Decision 2005/444/EC)
EU-CON. Classi ied In o ma ion: CONFIDENTIEL UE (Commission Decision 2005/444/EC)
EU-SEC. Classi ied In o ma ion: SECRET UE (Commission Decision 2005/444/EC)
D6.3 AI Fac o y Hub Handbook
4
Ve sion His o y
Re ision
Da e
Edi o s
Commen s
0.1
02.09.2025
Ee a Ha jula (CSC),
Ma in Duda (IT4I)
Fi s comple e d a o he documen eady
o e iew.
0.2
09.09.2025
Alan Medla (UH), Sa a
Ga a elli (CSC)
Re iew conduc ed.
0.3
16.09.2025
Ee a Ha jula (CSC)
Second d a comple ed and sen o he SMB
o inal commen s.
0.4
24.09.2025
Ee a Ha jula (CSC)
Final e sion a ailable o he PMO
1.0
25.09.2025
Anna Luoma (CSC)
Quali y check pe o med by he PMO and
sen o o icial e iew
Glossa y o Te ms
I em
Desc ip ion
AI Campus
The AI Campus is pa o he AI Fac o y Hub, se ing a pa icula a ge g oup,
s uden s specialising in o in e es ed in a i icial in elligence. I consis s o phys-
ical and i ual acili ies ha will i s be es ablished a he AI Fac o y Hub on
Aal o Uni e si y’s campus in O aniemi, Espoo, Finland. The AI campus and i s
ac i i ies and se ices p o ide a amewo k ha a ac s and nu u es alen ,
os e s in e ac i e communi ies, and o e s a ange o ac i i ies o suppo and
engage s uden s.
AI Fac o y Hub
The AI Fac o y Hub is he main en i y and he umb ella e m o he e en s,
aining, and esou ces ha he hub p o ides o i s di e en a ge g oups and
s akeholde s. I p o ides bo h physical and i ual co-wo king spaces ha ca e
o he di e se needs and goals o a ious s akeholde s, including s a ups,
SMEs, scien i ic communi ies, alen ed s uden s, HPC/AI suppo eams, incu-
ba o s, and accele a o s. The main si e will be on he Aal o Uni e si y campus
in Espoo, Finland which is a cen al hub o AI esea ch and coo dina ion in Fin-
land. The main si e will be suppo ed by subsidia y si es on o he campuses and
AI accele a o si es ac oss he conso ium pa ne s a es.
Digi al Decade
s a egy
Eu opean Union's 2030 Policy P og amme o achie e a human-cen e ed, sus-
ainable, and digi ally ans o med socie y by se ing conc e e a ge s o digi-
al skills, digi al in as uc u es, he digi aliza ion o businesses, and he digi al-
iza ion o public se ices. Ta ge s and u he in o ma ion he e.
Ellis Ins i u e Fin-
land
Ellis Ins i u e Finland is a newly es ablished wo ld-class esea ch hub in AI and
machine lea ning. I aims o de elop an ecosys em o op alen , businesses,
and in es men s in Finland, wi h undamen al machine lea ning esea ch a i s
co e. The Ins i u e is pa o he Eu opean Labo a o y o Lea ning and In elli-
D6.3 AI Fac o y Hub Handbook
5
gen Sys ems (ELLIS), which aims o build a pan-Eu opean ins i u ion o ma-
chine lea ning esea ch. The ins i u e is unde Aal o Uni e si y and is loca ed on
he uni e si y campus in O aniemi, Espoo. The O aniemi AI Fac o y Hub will be
es ablished in he same loo wi h Ellis Ins i u e. O icial websi e o Ellis Ins i u e
Finland he e.
HPC
High-Pe o mance Compu ing - supe compu ing esou ces o la ge-scale sim-
ula ions and modelling
HPDA
High-Pe o mance Da a Analy ics - in as uc u e o p ocessing massi e da-
ase s
LUMI AI Fac o y
The LUMI AI Fac o y is a pionee ing se ice in as uc u e and expe suppo
cen e designed o accele a e AI inno a ion ac oss Eu ope. I p o ides a seam-
less, open AI solu ion ha combines wo ld-class supe compu ing, high- alue
da a, and op- ie AI expe ise — empowe ing s a ups, SMEs, esea che s, and
businesses o de elop cu ing-edge AI models, ools, and applica ions. O icial
websi e he e.
LUMI AI Fac o y Se -
ice Cen e P ojec
(LUMI AIF)
P ojec ac onym LUMI AIF. The LUMI AI Fac o y Se ice Cen e p ojec is pa
o he b oade LUMI AI Fac o y ini ia i e ha in eg a es wo ld-class compu ing,
op alen and high- alue da ase s o accele a e AI inno a ion in Finland and
ac oss Eu ope. This ecosys em le e ages he cu en LUMI supe compu e bu
is gea ing up o he upcoming LUMI-AI sys em and he AI-op imised quan um
expe imen al pla o m LUMI-IQ. Mo e in o ma ion he e.
NAIC
The No wegian AI Cloud. NAIC p o ides access o High-Pe o mance Compu-
ing (HPC) esou ces and suppo o esea che s, PhD s uden s, educa o s, en-
ep eneu s, s a ups, and SMEs who a e in e es ed in explo ing AI me hodolo-
gies and de eloping ad anced AI models and algo i hms.
NRIS e-in as uc-
u e
No wegian Resea ch In as uc u e Se ices is a collabo a ion be ween Sigma2
AS and he ou uni e si ies NTNU, UiB, UiO og UiT, o o e esou ces and sup-
po p o iding high pe o mance compu ing and da a s o age se ices o e-
sea che s a ilia ed wi h academic ins i u ions in No way.
QC
Quan um Compu ing - eme ging quan um echnologies o specialized compu-
a ional p oblems
SMEs
Small and medium-sized en e p ises.
D6.3 AI Fac o y Hub Handbook
6
Execu i e Summa y
This handbook p o ides a s a egic and p ac ical guide o planning, es ablishing, and ope a ing AI Fac-
o y Hubs wi hin uni e si ies and esea ch en i onmen s. I is in ended o inspi e ins i u ions and s ake-
holde s by p esen ing a concep o wha an AI Fac o y Hub is, why i ma e s, and how i can be success-
ully implemen ed. The handbook o e s adap able amewo ks, ope a ional p inciples, and codes o con-
duc o suppo smoo h and sus ainable hub ope a ions.
D awing on he expe iences o he LUMI AI Fac o y Se ice Cen e p ojec (LUMI AIF), he handbook
highligh s bes p ac ices and lessons lea ned o guide new ini ia i es. The O aniemi Hub in Espoo, Fin-
land, which will be launched by he end o 2025, se es as he p ima y e e ence poin o he ne wo k o
he AI Fac o y Hubs. In addi ion, he Os a a Hub in Os a a, Czech Republic is cu en ly in de elopmen
and will be launched a a la e s age, u he expanding he AI Fac o y ne wo k. The plans o bo h hubs
ha e been conc e ized and p o ide a s ong ounda ion o u u e hubs. Ins i u ions a e encou aged o
build upon hese p ac ices while ailo ing hem o hei local ea u es, ne wo ks, and ecosys ems.
AI Fac o y Hubs a e posi ioned wi hin a wide ecosys em o na ional and Eu opean AI and HPC ini ia i es.
Unde s anding his s a egic landscape is essen ial o aligning wi h b oade goals, engaging ele an
s akeholde s, and maximizing impac . Each hub is encou aged o iden i y key ac o s in i s local ecosys-
em—such as uni e si ies, esea ch cen e s, companies, and public agencies—and ac i ely pu sue collab-
o a ion.
By sha ing p ac ices, pa icipa ing in join ini ia i es, and suppo ing c oss-bo de inno a ion, hubs con-
ibu e o he Eu opean AI Fac o y ision. Thei in eg a ion in o he LUMI AI Fac o y ecosys em and
alignmen wi h na ional and EU-le el s a egies posi ion hem as key d i e s o digi al ans o ma ion,
esea ch excellence, and capaci y building in AI and high-pe o mance compu ing.
D6.3 AI Fac o y Hub Handbook
7
Table o Con en s
1. In oduc ion ............................................................................................................... 9
1.2 Co e Value P oposi ion 10
2. The AI Fac o y Hub Concep ...................................................................................... 11
2.1 Vision and Mission 11
2.2 S a egic Objec i es and Ta ge Ou comes 12
2.3 Success Me ics and E alua ion 13
2.4 Code o Conduc 13
2.5 Gene al Layou and Facili ies 15
2.6 Roles and Responsibili ies 16
2.7 Engagemen and P og ams 17
2.8 Membe ship, Vi ual Collabo a ion and Communi y P ocesses 18
2.9 Go e nance and Feedback 20
2.10 Ecosys em Con ex and S a egic Posi ioning 20
2.11 In eg a ion wi h Na ional AI Ecosys ems and Se ice Cen es – Example: No way21
3. AI Fac o y Hub O aniemi Guide ................................................................................. 22
3.1 Hub O e iew 22
3.2 Con ac ing he AI Fac o y Hub 24
3.3 Visi ing he O aniemi AI Fac o y Hub 24
3.3.1 O e iew o O aniemi Hub isi o s 25
3.4 A ailable Se ices and Suppo o Hub Visi o s and Use s 25
3.4.1 Expe Suppo 25
3.4.2 T aining Ac i i ies and skills de elopmen 26
3.5 E en s and Engagemen 26
3.6 Code o Conduc 27
4. AI Hub Os a a Guide ............................................................................................... 28
4.1 Hub O e iew 28
D6.3 AI Fac o y Hub Handbook
8
4.2 Con ac ing he AI Fac o y Hub 28
4.3 Visi ing he Hub 29
4.3.1 O e iew o Os a a Hub isi o s 29
4.4 Use s, Guidelines, and Roles 29
4.4.1 Use G oups 29
4.4.2 AI Hub Coo dina o 29
4.5 A ailable Se ices and Suppo 29
4.6 E en s and Engagemen 30
4.7 Regional In eg a ion 30
4.8 Code o Conduc 30
D6.3 AI Fac o y Hub Handbook
9
1. In oduc ion
The pu pose o his handbook is o p o ide a comp ehensi e guide o es ablishing and ope a ing AI Fac-
o y Hubs wi hin uni e si ies and esea ch en i onmen s. The AI Fac o y Hub handbook is c ea ed o in-
s uc s a on p ac icali ies o he co-wo king spaces and s uden acili ies o ensu e smoo h ope a ion.
I also includes codes o conduc o he p emises. D awing on he expe iences o he LUMI AIF p ojec ,
he handbook ou lines p ac ical amewo ks, ope a ional p inciples, and implemen a ion s a egies ha
suppo he c ea ion and sus ainable ope a ion o such hubs. The c ea o s o his handbook wish ha i
could inspi e o ganisa ions in es ablishing new AI Fac o y Hubs.
The deli e able is s uc u ed in wo pa s. The i s in oduces he AI Fac o y Hub concep , while he sec-
ond o e s conc e e examples h ough he handbooks o he O aniemi and Os a a hubs. Among hese,
he O aniemi hub se es as he p ima y hub and e e ence poin o he LUMI AI Fac o y ne wo k, p o id-
ing he main ounda ion o p ac ices and guidelines o be adop ed and u he de eloped in subsequen
hubs. In addi ion, he e a e plans o es ablish new hubs a he Uni e si y o Helsinki’s Kumpula Campus
and o launch hub ac i i ies in Kajaani, Finland in collabo a ion wi h CSC’s local o ice and egional s ake-
holde s. Hubs es ablished a e O aniemi a e encou aged o adop and build upon he bes p ac ices
iden i ied he e, ensu ing consis ency and scalabili y ac oss he AI Fac o y Hub ne wo k, and bea ing he
local ea u es, ne wo ks and ecosys ems in mind.
This handbook b ings oge he insigh s om bo h Finnish and Czech implemen a ions, p o iding adap -
able guidelines ha acknowledge he di e si y o ins i u ional, cul u al, and egional con ex s in which
hubs ope a e. Ins ead o p esc ibing igid solu ions, i emphasizes co e p inciples and lexible p ac ices
ha can be ailo ed o local needs while main aining ope a ional excellence.
Key opics include guidelines o es ablishing a hub, as well as he essen ial ea u es de i ed om he AI
Fac o y Hub concep ha should be adap ed o new en i onmen s. Ins i u ions a e encou aged o apply
hese ecommenda ions wi hin hei speci ic con ex while upholding he sha ed alues o accessibili y,
excellence, and collabo a ion.
1.1 Ta ge Audiences and Use F amewo k
This handbook add esses mul iple s akeholde g oups ac oss di e en le els o engagemen , ha a e
ca ego ised in o h ee main g oups: p ima y ins i u ional s akeholde s, he hub use communi y and
seconda y s akeholde s. The ca ego isa ion ollows he p inciples o a s akeholde segmen a ion
amewo k, whe e di e en g oups a e iden i ied acco ding o hei ela ionship o a p ojec —in his
case, an AI hub.
Each hub-es ablishing o ganisa ion de e mines which s akeholde g oups a e essen ial o hei ope a-
ions, based on hei speci ic ole and hei connec ions wi hin local and na ional ne wo ks.
P ima y Ins i u ional S akeholde s
• Di ec ly in ol ed in planning, managing o unding he hub:
D6.3 AI Fac o y Hub Handbook
16
euse can con ibu e o bo h sus ainabili y and a sense o con inui y wi h he local con ex .
Accessibili y & Sus ainabili y:
Si e Selec ion: When selec ing p emises, conside p oximi y o key use g oups (s uden s, esea ch-
e s, indus y pa ne s), anspo connec ions, accessibili y o di e se use needs, and alignmen
wi h ins i u ional sus ainabili y and de elopmen s a egies.
• Ba ie - ee access o all spaces, including en ances, es ooms, and e en a eas.
• Clea signage and na iga ion suppo o new isi o s.
• In eg a ion wi h public anspo ou es and bike access o educe a el ba ie s.
• Ene gy-e icien ligh ing, hea ing, and cooling solu ions.
• Flexible layou s ha allow long- e m use and educe unnecessa y e u bishmen s.
2.6 Roles and Responsibili ies
A success ul AI Fac o y Hub ecognizes he di e si y o i s use base— om s uden s and esea che s o
indus y pa ne s and isi ing expe s. Unde s anding he speci ic needs, expec a ions, and le els o ex-
pe ience o hese g oups is essen ial o p o iding he igh kind o suppo , se ices, and engagemen
oppo uni ies.
Clea oles and esponsibili ies help ensu e smoo h ope a ions, a welcoming a mosphe e, and a con-
sis en use expe ience ac oss all hub ac i i ies. Depending on he hub’s phase (jus launched o es ab-
lished) and he numbe o pe manen s a , esponsibili ies and asks can be alloca ed as needed. Each
hub should de ine oles and esponsibili ies in a way ha bes sui s i s ope a ional con ex , including how
he Hub Manage and s a sha e esponsibili ies.
Hub Manage Responsibili ies
• Main ains a s a egic o e iew o he hub’s ope a ions, ensu ing ac i i ies a e aligned wi h i s
pu pose and ha he hub e ec i ely se es i s use s.
• Engages wi h s akeholde s and iden i ies new, ele an a ge g oups o expand he hub’s each
and impac .
• Rep esen s and p omo es he hub’s ope a ions and bes p ac ices o ele an s akeholde s.
• Is amilia wi h he hub’s ou ines, opening hou s, aining schedules, and se ices, and communi-
ca es hese o he co-wo ke s o ensu e smoo h ope a ions.
• Main ains o de and ensu es compliance wi h hub ules.
• Iden i ies esou ce needs and epo s ope a ional issues occu ing in he hub.
• Coo dina es he eedback collec ion p ocess and ensu es he eedback is p ocessed and imple-
men ed in u u e ac i i ies.
D6.3 AI Fac o y Hub Handbook
17
Suppo S a / Co-Wo ke s
• Visible and p esen a he on-si e hub, and he p esence should also be coo dina ed wi h he i -
ual hub o ensu e con inuous suppo .
• Welcome and guide use s o he app op ia e expe s o se ices.
• P o ide echnical o adminis a i e suppo .
• Assis wi h onboa ding new use s, including guidance on hub ac i i ies, p emises, guidelines and
a ailable suppo .
• Suppo e en s, wo kshops, and aining sessions.
• Regis e hub use s and collec ing eedback om hem.
• Repo ope a ional issues occu ing in he hub.
2.7 Engagemen and P og ams
AI Fac o y Hubs aim o c ea e inspi ing spaces o lea ning, collabo a ion, and inno a ion. To engage hub
use s and s akeholde s:
E en s and Ac i i ies
• Examples o e en s and ac i i ies: wo kshops, semina s, hacka hons, and ne wo king e en s.
• Ta ge g oups: s uden s, esea che s, indus y, and public s akeholde s.
• When planning e en s, conside audience, iming, capaci y, and accessibili y.
• Think abou wha kind o alue he e en s b ing o he a ge g oups and how hey se e he
hub’s ole, ision, and mission in he ecosys em.
• Ask he a ge g oups abou hei needs and wishes ega ding he e en s.
• By o e ing hyb id and emo e e en s, he e is a possibili y o each wide audiences and o se e
a ge g oups independen o ime and place.
Scheduling P inciples
• Annual o qua e ly planning o ensu e balanced p og amming.
• Keep he a ge g oups in mind when d a ing he annual plan.
• Flexibili y o espond o eme ging esea ch oppo uni ies o use needs.
• Conside a ge audience, echnical backg ound and accessibili y needs.
• U ilize app op ia e spaces based on e en scale and pu pose.
• Ensu e inclusi e pa icipa ion wi h language suppo when needed.
• Connec wi h egional inno a ion ecosys em.
• Collec pa icipan eedback o con inuous imp o emen .
P omo ion and Feedback
D6.3 AI Fac o y Hub Handbook
18
• Ensu e p og am isibili y and indabili y—whe he h ough a sha ed calenda , websi e, mailing
lis , o bulle in boa d—de eloped in collabo a ion wi h expe s in ol ed in aining and skills de-
elopmen .
• Collec eedback om pa icipan s on sa is ac ion, pa icipa ion, and imp o emen ideas.
• Use eedback o de elop new ac i i ies, adap schedules, and enhance se ices.
T aining and Lea ning Oppo uni ies
• Ta ge audiences: s uden s, esea che s, and p o essionals.
• Subjec a eas: HPC, AI me hods, domain-speci ic applica ions.
• Deli e y modes: on-si e, i ual, o hyb id o ma s depending on esou ces and use needs.
• T aining p og ams and expe imen a ion should be suppo ed by a secu ed and con olled en i-
onmen , ha enables sa e es ing and de elopmen o AI ools, models and wo k lows.
2.8 Membe ship, Vi ual Collabo a ion and Communi y P ocesses
The AI Fac o y Hub communi y includes bo h on-si e and i ual pa icipan s. To ensu e smoo h collab-
o a ion and a sa e, suppo i e en i onmen , he ollowing p ocesses apply ac oss all ac i i ies and pla -
o ms. These guidelines and ac ions can be adap ed based on he speci ic needs, ools, and con ex o
each hub. Flexibili y is encou aged o ensu e alignmen wi h local p ac ices and a ailable esou ces.
Vi ual Co-Wo king
• Remo e access o sha ed compu a ional esou ces may be p o ided when app op ia e.
• Collabo a ion ools ( ideo con e encing, p ojec managemen pla o ms, booking sys ems) sup-
po hyb id and online ac i i ies.
• The same Code o Conduc applies o all i ual en i onmen s as in physical spaces.
Access o HPC Resou ces
• When planning hub ea u es, communica ion and guidance ega ding access o HPC esou ces
should be aken in o conside a ion.
• The hub’s a ge g oups and use s should be conside ed. I is also essen ial o choose communi-
ca ion channels ha e ec i ely each po en ial hub use s.
• Communica e clea ly wi h hub use s and ensu e ha he coo dina o and/o hub s a can guide
isi o s on he gene al equi emen s ha mus be me o gain access o he esou ces.
• Walk-in basis du ing opening hou s in he on-si e hubs. Opening hou s o be de ined by he hos -
ing o ganisa ion.
Regis a ion and Hub Access
D6.3 AI Fac o y Hub Handbook
19
Please ake he ollowing in o accoun when planning and implemen ing hub p ac ices, whe he o he
on-si e p emises o he i ual hub en i onmen :
• Regis a ion equi ed o bo h i ual and on-si e hub isi o s and use s.
• All new membe s should e iew and ag ee o he Code o Conduc a he ime o he egis a ion.
• In o ma ion collec ed om hub isi o s and use s o p ojec epo ing and se ice imp o emen
includes name, a ilia ion, pu pose o isi .
• All ac i i ies will be conduc ed in acco dance wi h EU law o ensu e da a p o ec ion.
• O ien a ion: In oduc ion o hub esou ces, expec a ions, epo ing mechanisms, and suppo
channels. This can be conduc ed by he hub manage o ano he co-wo ke . A i ual e sion is
also possible, e.g. an in oduc ion ideo.
• Communi y Access: Membe s a e added o ele an collabo a ion pla o ms and mailing lis s.
• Con inuous Rein o cemen : Regula eminde s a e en s, upda es on policies, and e eshe
aining when needed.
Repo ing and Suppo
• On-si e: Issues can be aised di ec ly wi h he Hub Coo dina o .
• Online: Repo s can be made ia he designa ed email add ess o communi y channel.
U gen cases: Call building secu i y o he eme gency con ac p o ided.
• Anonymous: Repo s may also be submi ed h ough ins i u ional sys ems.
• Response: All epo s a e handled con iden ially, ai ly, and p omp ly, wi h suppo o e ed o
a ec ed indi iduals.
En o cemen
• P og essi e Response:
o Ve bal o w i en wa ning.
o Tempo a y suspension o access.
o Exclusion om speci ic ac i i ies.
o Pe manen ban om hub se ices.
o O ganize s may ac immedia ely in se ious cases o ensu e communi y sa e y.
• Appeals can be submi ed o designa ed go e nance o leade ship con ac s.
O boa ding
• Volun a y Depa u e:
o Re u n any bo owed equipmen o access ca ds.
o Documen ongoing p ojec s and hand o e esponsibili ies.
o P o ide eedback h ough an exi in e iew (op ional).
o Op ion o emain connec ed o he communi y ne wo k as an alumni (op ional).
• In olun a y Remo al:
o Immedia e e oca ion o access in cases o se ious iola ion.
o Clea communica ion o easons and documen a ion o he inciden .
D6.3 AI Fac o y Hub Handbook
20
o Possibili y o eins a emen a e e iew on a case-by-case basis.
2.9 Go e nance and Feedback
E ec i e go e nance and con inuous eedback a e essen ial o main aining a well- unc ioning AI Fac-
o y Hub. While go e nance also includes aspec s such as unding, s a egic planning, and o e all man-
agemen , his sec ion ocuses on he p ac ical ope a ional p ocedu es ha suppo day- o-day decision-
making and use engagemen .
This sec ion complemen s Sec ion 2.3 (Success Me ics and E alua ion) by ou lining how hubs can imple-
men anspa en s uc u es and eedback loops o ensu e esponsi eness, accoun abili y, and con inu-
ous imp o emen .
Hubs should es ablish a clea and anspa en go e nance s uc u e, along wi h egula eedback mech-
anisms, including:
• Decision-making p ocedu es o ope a ions, e en s, and esou ce alloca ion.
• Con inuous eedback collec ion om use s and isi o s h ough su eys, in o mal discussions, o
digi al channels.
• Feedback-d i en imp o emen s o hub layou , se ices, and ac i i ies.
• Moni o ing o success me ics, such as use numbe s, sa is ac ion le els, and esea ch ou pu s.
2.10 Ecosys em Con ex and S a egic Posi ioning
AI Fac o y Hubs ope a e wi hin a b oade ecosys em o na ional and Eu opean AI and HPC ini ia i es.
Unde s anding his con ex is essen ial o s a egic alignmen , s akeholde engagemen , and maximiz-
ing impac . Each hub should iden i y key o ganiza ions and ac o s in i s local ecosys em—such as uni e -
si ies, esea ch cen e s, companies, and public agencies—and ac i ely seek collabo a ion oppo uni ies.
These pa ne ships no only s eng hen he hub’s local ele ance bu also con ibu e o he b oade Eu-
opean AI Fac o y ision.
By sha ing p ac ices, pa icipa ing in join ini ia i es, and suppo ing c oss-bo de inno a ion, hubs play
a i al ole in shaping a connec ed and o wa d-looking AI ecosys em. Recognizing he hub’s posi ion
wi hin his landscape helps guide decision-making, align wi h s a egic objec i es, and ensu e meaning-
ul con ibu ions a bo h na ional and Eu opean le els.
Key conside a ions include:
• The hub’s ole wi hin he LUMI AI Fac o y ecosys em.
• I s connec ion o Eu opean and na ional AI/HPC ini ia i es.
• Key s akeholde s and o ganiza ions o engage wi h (e.g., uni e si ies, esea ch ins i u es, com-
panies, public sec o ).
D6.3 AI Fac o y Hub Handbook
21
• The impo ance o unde s anding he hub’s posi ion in he b oade landscape o maximize im-
pac and collabo a ion.
• Re e ence o o he chap e s (e.g., go e nance, engagemen , s a egic objec i es).
2.11 In eg a ion wi h Na ional AI Ecosys ems and Se ice Cen es –
Example: No way
The pu pose o his sec ion is o p o ide an example o a po en ial u u e in eg a ion wi h a na ional AI
ecosys em. I has been ecognised in he LUMI AIF ha he success o he AI Fac o y Hubs also depends
on hei abili y o align wi h and ex end na ional AI ecosys ems. Na ional se ice cen e s ac as he en y
poin s o esea che s, public adminis a ion, and indus y, and ensu e ha in es men s a he Eu opean
le el a e ancho ed in domes ic p io i ies.
No way p o ides a conc e e example h ough he No wegian AI Fac o y, which is an ecosys em and se -
ice cen e , coo dina ed by Sigma2 and closely linked wi h he No wegian AI Cloud (NAIC) and NRIS e-
in as uc u e. This na ional ecosys em is embedded wi hin he LUMI AIF, and i s ole is o ensu e seam-
less in e ope abili y be ween No wegian use s and he wide AI Fac o y Hub ne wo k.
The in eg a ion ope a es on wo ecip ocal axes. Fi s ly, i makes No wegian NAIC se ices accessible ia
he hub. This includes AI- eady i ual machine images, model caching and sha ing pla o ms, and an AI-
HPC aining en i onmen o esea che s, IT p o essionals and s uden s. Addi ionally, i p o ides bench-
ma king se ices, ools o assessing AI wo kloads, and us ed da a en i onmen s o handling sensi i e
da ase s.
Secondly, he in eg a ion enables No wegian use s o access hub-le el se ices. Resea che s, SMEs, and
public sec o ac o s in No way gain s eamlined access o LUMI AIF esou ces, such as la ge-scale model
aining, ede a ed lea ning amewo ks, and pan-Eu opean da ase s. This is acili a ed ia ha monised
use po als, aligned alloca ion p ocesses, and in e ope able APIs, ensu ing ha use s can mo e seam-
lessly be ween na ional and hub-le el se ices.
To sus ain his bidi ec ional in eg a ion, LUMI AIF and he No wegian se ice cen e coo dina e in h ee
a eas:
• Technical in e ope abili y: common s anda ds o au hen ica ion, da a access, and con aine -
ised wo k lows o ensu e po abili y and compliance.
• T aining and skills: co-b anded wo kshops, hacka hons, and cu icula ha b idge na ional ain-
ing wi h LUMI AIF’s pan-Eu opean p og amme.
• Go e nance: egula exchange be ween LUMI AIF hub leade ship and na ional se ice cen e
ep esen a i es o align p io i ies, a oid duplica ion, and ensu e ecip ocal bene i .
This model shows how a na ional ecosys em, ancho ed in exis ing in as uc u es and compe ences, can
bo h en ich and bene i om he AI Fac o y Hub ne wo k. I p o ides a bluep in o o he conso ium
membe s o s uc u e hei na ional–Eu opean in eg a ion.
D6.3 AI Fac o y Hub Handbook
22
3. AI Fac o y Hub O aniemi Guide
This chap e in oduces he AI Fac o y Hub O aniemi, a collabo a i e space designed o suppo s uden s,
esea che s, and p o essionals wo king in he ield o a i icial in elligence. I p o ides p ac ical guidance
o hub isi o s and use s on:
• Wha he hub o e s.
• Who can use he acili ies.
• How o ope a e wi hin he space.
3.1 Hub O e iew
Loca ed on he 4 h loo o Maa in ie 8 in Espoo—on he same loo as he Ellis Ins i u e Finland, a wo ld-
class AI esea ch cen e — he AI Fac o y Hub O aniemi o e s a e sa ile and engaging en i onmen . I
includes indi idual wo k desks, mee ing ooms, and open spaces o aining, e en s and in o mal ga h-
e ings, os e ing a ib an and collabo a i e a mosphe e.
The hub is in ended o s uden s, esea che s, expe s, company ep esen a i es, and o he s akeholde s
connec ed o he LUMI AI Fac o y ne wo k. I suppo s bo h indi idual wo k and communi y-building
h ough i s se ices and suppo , wo kshops, e en s, and aining sessions. A i ual hub will also be es-
ablished, enabling emo e pa icipa ion ega dless o he physical loca ion.
The AI Campus, as pa o he AI Fac o y Hub, ocuses on engaging s uden s who a e in e es ed in AI and
HPC. Th ough i s ac i i ies and se ices, he campus builds a amewo k ha a ac s and nu u es alen ,
os e s in e ac i e communi ies, and o e s a ange o oppo uni ies o suppo and engage s uden s.
Loca ion: AI Fac o y Hub O aniemi, 4 h loo , Maa in ie 8, Espoo, Finland
En i onmen
• ~121 m² dedica ed o he AI Fac o y Hub including indi idual wo k desks and open space o ain-
ings and e en s.
• Mee ing ooms can be booked om he Ellis Ins i u e p emises on he same loo .
• Addi ional p emises on he 4 h loo : Sauna and e ace a ailable upon eques o special occa-
sions.
• P io i y sys em based on use classi ica ion and p ojec needs.
• Online booking sys em o all spaces and equipmen .
Ta ge Use s
D6.3 AI Fac o y Hub Handbook
23
O aniemi AI Fac o y Hub welcomes indi iduals, such as s uden s and expe s, as well as membe s o com-
panies and o ganisa ions in he LUMI AI Fac o y ecosys em.
• S uden s - Whe he beginning a bachelo ’s, pu suing a PhD, o anywhe e in be ween, he hubs
o e access o ools, aining, and a communi y o g ow he skills.
• Resea che s and academic s a – Oppo uni ies o s eng hen p ojec s wi h ad anced compu-
ing powe and expe suppo .
• S a ups and companies - om small eams o SMEs and la ge en e p ises, he hub p o ides pos-
sibili ies o es and scale AI ideas wi h wo ld-class esou ces.
• Go e nmen o ganisa ions and public sec o – Oppo uni ies o come up wi h solu ions ha
bene i socie y.
Special Fea u es
• Access o HPC, QC and LUMI AI Fac o y esou ces unde ce ain condi ions and e ms. These will
be cla i ied and communica ed o hub use s as soon as he egis a ion sys ems and esou ce
alloca ions ha e been de ined. This is expec ed o happen by he end o 2025.
• Expe suppo and guidance. Examples include he possibili y o book a ime slo o consul ing
a LUMI AIF expe and he access o he sel -lea ning ma e ials. Fu he ins uc ion o be upda ed
by he end o 2025.
• AI Campus p og ams o s uden s in e es ed in a i icial in elligence, including aining and wo k-
shops.
• In e ac ion and ma chmaking wi h indus y.
Figu e 1 Maa in ie 8 om ou side
D6.3 AI Fac o y Hub Handbook
24
3.2 Con ac ing he AI Fac o y Hub
Gene al Con ac
• Gene al Inqui ies: Un il he launch, ee a.ha jula@csc. i o lumi-ai -wp6@pos i .csc. i
• Hub Coo dina o : [Con ac o be upda ed upon ec ui men ]
• P ojec Leade ship: Ee a Ha jula (CSC), p ojec manage o AI Hubs; Juhani Hu unen (CSC),
Lead o LUMI AI Fac o y Ecosys em De elopmen
• AI Campus Ac i i ies: [email o be es ablished]
Campus Maps, Public T anspo , Pa king and Opening Hou s: Fu he in o ma ion abou he Maa in ie
8 building and ge ing a ound in O aniemi Campus.
Figu e 2 Maa in ie 8 on he map
Physical Accessibili y
• Building ecep ion on he 1s loo in he main lobby o he Maa in ie 8 building.
• Visi o egis a ion equi ed a ecep ion.
• Building equipped o accessibili y needs. Please con ac cus odians a uas-
ah imes a i @aal o. i o +358504723222.
3.3 Visi ing he O aniemi AI Fac o y Hub
The de ails ega ding isi s will be de ined and communica ed o hub isi o s by he launch. In he isi ing
guidelines, he p inciples om sec ion 2.8 — Membe ship, Vi ual Collabo a ion and Communi y P o-
cesses — will be conside ed, applied o he O aniemi Hub p ac ices, and communica ed o i s isi o s and
use s.
D6.3 AI Fac o y Hub Handbook
25
3.3.1 O e iew o O aniemi Hub isi o s
• Gene al opening hou s om Monday o F iday, 8:00 AM – 17:00 PM.
• Ex ended access a ailable o egis e ed use s and special p og ams. Flexible scheduling o
e en s and wo kshops.
• Dedica ed ime slo s o speci ic a ge g oups, such as expe suppo se ices, may be in o-
duced in he u u e. De ails will be announced once he scheduling and o ma s a e con i med.
• Simple egis a ion o p ojec epo ing and se ice imp o emen is equi ed om all hub isi-
o s and use s.
○ In o ma ion collec ed: name, a ilia ion, pu pose o isi .
3.4 A ailable Se ices and Suppo o Hub Visi o s and Use s
3.4.1 Expe Suppo
The O aniemi Hub aims o sp ead knowledge and os e inno a ion wi hin bo h he hub ne wo k and he
b oade LUMI AI Fac o y ecosys em. Regis e ed use s engaged in AI- ela ed skill de elopmen , s udies
o RDI ( esea ch, de elopmen and inno a ion) can access selec ed esou ces and se ices p o ided by
he LUMI AI Fac o y. These include, o example, compu ing capaci y, da a se ices and AI compu ing
ools. The ollowing sec ions ca ego ize he planned se ices and suppo and desc ibe how he hub
wan s o engage wi h i s isi o s.
AI Compu ing Capaci y
• Di e en le el o compu ing packages a e o e ed o a ious needs: packages sui able o ain-
ing con en ional machine lea ning models, ine uning la ge language models, o combining
quan um compu ing and AI o HPC compu ing.
Access o Da a
• Sui able solu ions o di e en demands: Da ase s-as-a-Se ice, s o age o la ge da ase s, s o -
age o sensi i e da ase , suppo o loca ing and acqui ing new da a o men ion some exam-
ples.
AI Compu ing Se ices
• AI model se ice and packages o cloud in as uc u e o da a s o age.
Suppo o AI Adop ion in Companies
• Fo indus y ep esen a i es he hub o e s analysis and epo o iden i y he needs and ecom-
mended se ices.
• Suppo o s a ing la ge scale AI compu ing o o applying compu ing o da a esou ces
