Machine Learning Engineer

JobOverview

TheDataScientist/MLEngineerbuildsanddeployspredictivemodelsandanalyticalsystemsthatturnAGS'splayerandgamedataintoquantitativeinsightsthatdirectlyimprovegamedesignandcommercialdecisions.Thisrolebridgesbehavioraldatascience(understandinghowplayersinteractwithgames)andproductionMLengineering(deployingmodelsthatactuallyreachdecision-makers).Itfeedsgamedesignerswithdata-drivendesignrecommendationsfortheML-drivengamedesigninitiative,supportsyieldmanagementwithpredictivemodelsforInteractiveYieldMax,andenablesoperatorstounderstandtheirplayerbasemoredeeplyanchoredtoAGS'sTech&DataheromissionofanaccessibledatalayerwithliveKPIspoweringeverydecision.

Responsibilities

• Buildplayersessionbehavioralmodelsretentionprediction,abandonmentmodeling,post-bonusbehavioranalysis,andbetescalationmodelingfromiGamingsessiondata
• DevelopgameperformancepredictionmodelspredictWPUPD,timeondevice,andfloorlongevityfromgamespecificationfeaturesandhistoricalperformancedata,usingagamefeatureextractionpipelinethatreverse-engineersexistingtitlesintostructured,reusablefeatures
• Buildmathmodeloptimizationanalyticsanalyzeactualvs.theoreticalRTP,hitfrequency,andbonusfrequency;identifymathmodelanomaliesacrossthedeployedfleet
• Createplayersegmentationmodelsclusterplayersintobehavioralarchetypes(bonushunters,jackpotchasers,basegamegrinders)toinformgamedesignandoperatorrecommendations
• SupporttheInteractiveYieldMaxyield-managementtoolbuildtheunderlyingmodelsthatpredictwhichAGSgamemaximizesperformanceinagivenfloorposition,operatorproperty,andplayerdemographic
• Buildpredictivemaintenancemodelsanalyzecabineterrorlogsand,assensor/telemetrypipelinesmature(DynamicsFieldService/Dataverse),incorporatetelemetrytoidentifyfailureprecursorpatternsandpredictcomponentfailures
• Feedgamedesigndecisionstranslatemodeloutputsintogamedesigner-friendlyinsightsthatareactionableinthegamespecificationprocess
• DesignandanalyzeA/Btestsexperimentaldesign,statisticalanalysis,andresultsinterpretationforgamemathvarianttesting(whereregulatorilypermitted)
• ProductionalizemodelspackagemodelsfordeploymentonAzureML/Fabric,withMLflow-basedregistry,monitoring,andretrainingpipelines

Skills/Requirements

• 48 years of data science and/or ML engineering experience, with demonstrated production model deployment (not just notebook analysis)
• Behavioralanalyticsexpertisehasbuiltretention,churn,orengagementmodelsusingevent-levelbehavioraldata(sessionlogs,clickstreams,transactionsequences)
• StrongPythonandSQLskillspandas,scikit-learn,XGBoost,statsmodels;canquerythedatawarehouseindependently(amixofon-premSQLServerandSalesforcetoday,migratingtoMicrosoftFabric/OneLake)withoutrelyingonadataengineerforeveryanalysis
• Statisticalrigorsurvivalanalysis,A/Btestdesign,causalinference,regressionmodeling;understandsthedifferencebetweencorrelationandcausation
• Machinelearningbreadthclassification,regression,clustering,recommendationsystems;canselecttherightmodelingapproachforeachproblem
• Datacommunicationskillscantranslatemodeloutputsintobusiness-friendlylanguagethatgamedesignersandcommercialleaderscanacton
• Experiencewithmessy,real-worlddatacomfortablewheregamefeaturesaren'tfullydocumentedandpipelinesarestillbeingbuilt;doesn'trequireperfectdatatodelivervalue
• Bachelor'sorMaster'sdegreeinDataScience,Statistics,ComputerScience,Mathematics,orrelatedquantitativefield

Preferred

• Gaming,mobilegaming,orconsumerbehavioralanalyticsexperience
• FamiliaritywithcasinogamemechanicsRTP,volatility,Hold&Spin,theoindex
• ExperiencewithtimeseriesanalysisandanomalydetectionforIoT/sensordata
• Knowledgeofresponsiblegamblingdataconsiderations
• Experience with MLflow, Azure ML, or Fabric Notebooks/Spark for model lifecycle management

Note:Alloffersarecontingentuponsuccessfulcompletionofabackgroundcheck

*Postedpositionsarenotopentothirdpartyrecruitersandunsolicitedresumesubmissionswillbeconsideredfreereferrals.

AGSisanequalopportunityemployer