在 Asana 构建企业级 ML 体验的见解.pdf

《在 Asana 构建企业级 ML 体验的见解.pdf》由会员分享，可在线阅读，更多相关《在 Asana 构建企业级 ML 体验的见解.pdf（36页珍藏版）》请在三个皮匠报告上搜索。

1、AKO|FY24Databricks Data+AI Summit 202411 June 2024Insights from building enterprise-grade ML Experiences at Asana2Sarang GuptaML Data ScientistML Infrastructure EngineerZhan ShiAbout us 3123AgendaIntroductionProject Recommendation SystemAsana and the Work Graph AI&ML Themes and CapabilitiesKey Insig

2、hts and LearningsProductionizationML Infrastructure4Chapter 1IntroductionAsana and The Work GraphITOperationsMarketingSalesProductPowered by theAsana Work GraphTeamsTasksProjectsITOperationsOperationsMarketingSalesProductPowered by theAsana Work GraphTeamsTasksProjectsPortfoliosGoalsTasksProjectsPor

3、tfoliosITOperationsITOperationsMarketingSalesProductPowered by theAsana Work GraphTeamsAIAutomationReportingResource ManagementWorkflowsIntegrationsSecurityGoalsTasksMissionITOperationsITOperationsTeamsMarketingSalesProductProjectsPortfoliosData811 membersComms6,183 membersIntegrity632 membersGrowth

4、95 membersR&D5,041 membersAIAutomationReportingResource ManagementWorkflowsIntegrationsSecurityAIAutomationReportingResource ManagementWorkflowsIntegrationsSecurityPowered by The Asana Work Graph Data ModelThe Asana Work Graph61.Ranking2.Classification3.Detecting anomalies4.Recommendations5.Predicti

5、ng future behavior6.Extracting meaning from text7.Hypothesization8.Intent unwinding9.Conversing in text10.Summarization/translation11.Reasoning and making decisions12.Reasoning and acting13.Learning and durable memory14.Chaining and composabilityPattern recognitionRecommendation and predictionUnders

6、tanding and using natural languageReasoning,learning and autonomyAI/ML at AsanaCapabilities and themesML/DL on structured dataLLMs7Smart RankingsHelps users to navigate to their work easilyProject RankingsFacepile RankingsPriority Tasks in EmailsWe use smart ranking to help our users to navigate bet