人工智能论文英文版-MLOps with Microservices：A Case Study on the Maritime Domain.pdfVIP

MLOpswithMicroservices:

ACaseStudyontheMaritimeDomain

RenatoCordeiroFerreira1,2,3[0000−0001−7296−7091],

RowanneTrapmann1,2,3[0009−0000−1851−7508],and

Willem-JanvandenHeuvel1,2,3[0000−0003−2929−413X]

1JheronimusAcademyofDataScience(JADS),’s-Hertogenbosch,TheNetherlands

52EindhovenUniversityofTechnology(TUe),Eindhoven,TheNetherlands

23

0TilburgUniversity(TiU),Tilburg,TheNetherlands

2{r.cordeiro.ferreira,w.j.a.m.v.d.heuvel}@jads.nl

r.i.a.h.trapmann@tue.nl

n

u

J

6

Abstract.Thiscasestudydescribeschallengesandlessonslearnedon

]buildingOceanGuard:aMachineLearning–EnabledSystem(MLES)

Eforanomalydetectioninthemaritimedomain.First,thepaperpresents

Sthesystem’sspecification,andarchitecture.OceanGuardwasdesigned

.withamicroservices’architecturetoenablemultipleteamstoworkon

s

ctheprojectinparallel.Then,thepaperdiscusseshowthedevelopers

[adaptedcontract-baseddesigntoMLOpsforachievingthatgoal.Asa

MLES,OceanGuardemployscode,model,anddatacontractstoes-

1tablishguidelinesbetweenitsservices.Thiscasestudyhopestoinspire

v

2