1、Powering real-time market insight with Databricks AppsHuy NguyenDAIS 2025AgendaOur dashboard journeyIngestion optimisationDashboard optimisationIntroductionIntroductionWho is Optiver?Global liquidity provider and market makerTrading financial instruments globally on more than100 exchangesMultiple as
2、set classes IntroductionWhat is market making?Continuous quoting on the bid and offerProviding prices in all market conditionsWe improve the market by-Providing liquidity-Enabling price discovery-Improving transparency and efficiencyIntroductionMaking data-driven decisionsWhere is the next business
3、opportunity?Which trading strategies are optimal?Are we compliant with all regulations?Real-time insightsStrategic insightsRegulatory complianceIntroductionDatabricks at Optiver500+Monthly active users4000+Data sourcesPBsTotal volumeIntroductionDatabricks at OptiverQuantitative ResearchLarge scale d
4、ata processingMLDashboardsOptiversDashboard JourneyTwo years ago All data was on-prem Non centralised data sources:multiple databases,multiple data centers Non-analytical databases High friction to productionise dashboard-Researchers code-C+-Building new dashboards via GUI or DashThe on-prem struggl
5、eDB1DB2DB3Data Center AData Center BDB1DB2DB3Data Center CDB1DB2DB3One year ago Onboarded Databricks Explosion of AI/BI dashboard uses Easy access to AI/BI dashboard led to the democratisation of dashboarding Little supported required from dev teams Queries that were thought to be impossible became
6、trivialDashboard revolutionEight months ago AI/BI dashboards started being used for live use cases Overcrowding of warehouses due to constant refreshing Lack of version control Front end performance hit from rendering large amount of dataVictim of successSix months agoFinding a new path Trading team