apidays LIVE Hong Kong 2021 - Building AI/ML model API for Digital App by Eason Lai, AIA
0 likes651 views
The document outlines the development and application of AI/ML models and APIs by Eason Lai, a lead digital solution architect at AIA Hong Kong & Macau. It includes information on data processing, predictive analytics, and optimizing business processes, emphasizing integration with various data sources and technologies. Additionally, it highlights typical workflows and usage patterns in deploying AI/ML solutions for real-time insights and operational improvements.
apidays LIVE Hong Kong 2021 - Building AI/ML model API for Digital App by Eason Lai, AIA
- 1. AIA confidential and proprietary information. Not for distribution. AIA confidential and proprietary information. Not for distribution. Eason Lai Lead Digital Solution Architect, Digitalization & Innovation AIA Hong Kong & Macau Medium @ https://easonlai888.medium.com/ GitHub @ https://github.com/easonlai Building AI/ML model API for Digital App
- 2. AIA confidential and proprietary information. Not for distribution. AI ML Maturity Raw Data Cleaned Data Standard Reports Ad Hoc Reports & OLAP Generic Predictive Analytics What happened? Why did it happen? What will happen? Predictive Modeling What is the best that could happen? Optimization PREDICT & ACT SENSE & RESPOND AI ML Maturity Competitive Advantage
- 3. AIA confidential and proprietary information. Not for distribution. 3
- 4. AIA confidential and proprietary information. Not for distribution. 4 Common Usage Patterns of AI ML API Improve Existing Systems, Processes Business Process Optimization Reporting Predictive Maintenance Accelerate Business Forecasting Sales, Revenue Demand and Supply Pricing Deliver Results Quickly Real-time Insights Recommendations Fraud Detection
- 5. AIA confidential and proprietary information. Not for distribution. Prepare Experiment Deploy Typical E2E Flow of Building AI ML API Feedback Loop
- 6. AIA confidential and proprietary information. Not for distribution. 6 API Developers Data Scientists IDE Notebook Git Repos CI/CD Tool Model API Containers In Kubernetes Model Repos Containers Registry 1 2 1 3 2 1 2 3 4 Delivery Flow of AI ML API
- 7. MODEL & SERVE INGEST STORE PREP, ANALYTIC & TRAIN Sensors, IoT, Digital Apps (unstructured, semi-structured) Apache Kafka Mongo DB Files (unstructured, semi-structured) Media (unstructured) Logs (unstructured, semi-structured) HDFS Apache Spark Real-time Digital Apps Business/custom apps (structured) PostgreSQL BI & Report Apache Spark Real Time AI/ML Digital Apps