高效 MLOPS：使用 DATABRICKS 开发和部署 ML 模型（重复）.pdf

1、2024 Databricks Inc.All rights reservedLavinia GuadagnoloLavinia GuadagnoloAlessandro MazzulloAlessandro Mazzullo1Efficient MLOps:Developing and Deploying ML Models with Databricks2024 Databricks Inc.All rights reserved2024 Databricks Inc.All rights reserved2Who is this talk for?Who is this talk for

2、?Data scientistsData engineersData analystsML EngineersOrganizations2024 Databricks Inc.All rights reservedAgenda:Agenda:3About usChallenges and motivationsMain toolsUnity CatalogDatabricks Asset BundlesMLOps for MLHopesCI/CD Demo2024 Databricks Inc.All rights reserved2024 Databricks Inc.All rights

3、reserved4About About usus2024 Databricks Inc.All rights reserved2024 Databricks Inc.All rights reserved5About usAbout usProduct Owner,Data ScientistPlenitudeLavinia GuadagnoloLavinia GuadagnoloProduct Owner,Data ScientistPlenitudeAlessandro MazzulloAlessandro MazzulloMLHopes2024 Databricks Inc.All r

4、ights reserved2024 Databricks Inc.All rights reserved6About usAbout usWe are an international BenefitBenefit CompanyCompanyoperating in the energyenergy sectorsector since 2017We are active in the retailretail market offeringenergy and related services to more than1010 mlnmln customerscustomersWe op

5、erate in energy production fromrenewablerenewable sourcessources and in the world of e e-mobilitymobilityOur approach is focused on sustainabilitysustainabilityand innovationinnovation2024 Databricks Inc.All rights reserved2024 Databricks Inc.All rights reserved7ChallengesChallenges2024 Databricks I

6、nc.All rights reserved2024 Databricks Inc.All rights reserved8Motivation&ChallengesMotivation&ChallengesAgilityAgilityMManualanual steps for industrialisationDifficulty in bug bug fixingfixingScalabilityScalabilityNeed for easy integration for the many new modelsnew modelsMultiplication of code stru