Data and AI：Past, Present and the Future-堵俊平

1 .Data & AI: Past, Present and Future 堵俊平 Datastrato Founder & CEO ASF Member, Ex-Chair of LF AI & DATA 1 

2 .Data and AI: two sides of the same coin Data is consistently hot for over decades AI's current boom is the largest tech wave in decades Refer: trends.google.com 2 

3 .We are in Big-Waves of Data and AI Key Drivers: Buzzword: Internet, Social Media, Communication, Digital Machine Learning, Deep Learning, AI Framework, Photos, Services, Internet of Things, Transformer, AIGC, LLM, AGI, etc. Metaverse, Generative AI Refer:www.statista.com/statistics/871513/worldwide-data-created/ Refer: LifeArchitect.ai/models 3 

4 .Data Evolution with so many choices - open source or not RedShift SQL BigQuery Snowflake Databricks Starburst Kinesis Dataflow Confluent Streaming Analytical OLAP Impala Hive Spark Doris Flink Kafka Redpanda � � � Open Source � � � OpenAPI Governance Data Lake Storage Delta Lake Iceberg Hudi HMS Ranger Atlas Hadoop Ozone HBase Lake Azure GCP Data Lake Glue AD Catalog S3 Blob Storage GCS Formation Data Lake 4 

5 .Model Evolution: from Training to Ecosystem GPT LLaMA Refer: 2023 state of Data + AI by Databricks 5 

6 . Quantity Data is underwater Quality Discovery Governanc e 

7 .New Silos for Data and AI Ref: 2023 snowflake summit https://a16z.com/emerging-architectures-for-modern-data-infrastructure/ Multi-region data and model for business globally “Different Dialect” between Data and AI 7 

8 .Rethink & Reinvent - Multi-cloud “Data Stratosphere” SSOT for “Meta” Smart Governance Nature Language Interface Predictable Optimizations Codeless and Shortening ETL, less across the cloud 8 

9 .Synergy between DATA and AI Improvement of data-driven AI model Representation Accuracy Integrity Consistency Availability Data Data Data Insight Mining cleansing AI models suggest the ability of data insight 9 

10 .Project Gravitino: Break Down the Silos Building Data Fabric across the orgs Power AI to analyze the data intelligently Gravitino 

11 .Revolution Way to Build a Unified Metadata Lake Gravitino Unified Metadata Lake Managemen Optimizing Monitoring Auditing t Gravitino Data Message Data Lake Files RDBMS Model Warehouse Queue Gravitino Gravitino Key features Geo- AI + Data Catalog + Security in SSOT distributed Catalog Governance One place Arch Gravitino will be open source soon, welcome early adopters For more details, please attend afternoon session 11 

12 .We are hiring！ hr@datastrato.com 12 

13 .13