Built firm level machine-learning && AI platform for all LOBs and Data Scientists

* Lead a team of 5 to design and develop AI Platform in onboarding with complicated distributed transactions, including multiple regions, batch/individual onboarding, which currently supports around 300 projects and 2500 active users

* Lead and implement overall design to seamlessly integrated our on-premise infrastructure (cloud, MariaDB) with open cloud provider(AWS, DynamoDB) which successfully made onboarding process smoothly in view of bank regulatory security and compliment requirements.

* Highly get involved with AWS Machine Learning related service to work with Jupyter Notebook, including SageMaker, Comprehend, Transcribe, EMR, Glue, Athena, NAT etc

* Researched and Adopted Ray as distributed execution framework to model training which reducing 70% time on large model using cluster of nodes.

* Lead and design bunch of auxiliary micro services, Backfill service, Repaving service and Janitor service

* Integrated with in-house security module to handle security in both project and user level, support both OAuth and SAML in both of backend and UI Service

* Developed and set up monitor metrics usage/failure rate of notebook and using Grafana and tracking system using Splunk with CloudWatch,

* Actively work on full life cycle of machine learning platform, including training /serving platform, ML python SDK, model repository/promotion, etc

* Currently leading Migration and Refactor whole flow and infrastructure into open cloud Aws and adopt terraform to provision aws resources

• Build Multiple SpringBoot Based RESTful API Microservices from scratch to production in cloud AWS

• Highly get involved most of AWS during development and deployment, EKS, ECS, EC2, ECR, Route53, S3, DynamoDB, SES, etc

• Implement bunch of both OIDC and SAML applications In OKTA to support authentication and authorization to support our multi-tenancy services

• Create and set up of ELK (Elasticsearch Logstash/filebeat Kibana) to centrally store and monitor logs for different tenancies.

• Write Kotlin language to dynamically set up CI/CD pipeline for tons of projects in separately environment with Docker's help in Teamcity

• Heavily use kubernetes to deploy cluster environments and terraform to create infrastructure

• Get involved to create React UI application supported by backend microservices

• Fully migrate applications, users, groups data from OKTA to internal App Keti which is a wrapper of OKTA.

• Skillset used: SpringBoot, Java, AWS, K8s, ELK, Docker, Kotlin, Swagger, OKTA, Python, GO, React, Gradle, Terraform, Teamcity, Github, Bintray, CircleCI.

• Developed and implemented java code based automated testing framework for both online flights shopping UI and sabre internal JSON Restful Web service with JsonClient, Selenium, web driver.

• Leaded design and implemented javascript code based automated Web UI product testing framework with Gulp, Chimp and Node.js used in multiple browser.

• Reduced testing time 40% by dynamically exporting data from service calls/loading files to randomly generate the test data and refactoring structure and code with high reusability and efficiency across multiple teams.

• Performed the full cycle of testing including Unit Testing, Feature Function Testing, Integration Testing, customer validate testing and successfully deliver sabre products to many airline customers including Etihad Airways, Aero Mexico, Alitalia Airlines, Air Serbia, Bangkok airlines.

• Set up Jenkin job and make configuration management in JConsole on sabre internal Tomcat server to make build successfully compiled and run sanity test on delivered web service to make testing environment clean and runnable.

• Defect reporting/tracking and final test results reports creation and the tools JIRA, Rally, HP quality center etc.

• Used Spring Framework and Java Decorator API provided by EIT FedEx Development Services to publish and subscribe to the JMS messages infrastructure (including 2 publishers and 6 subscribers).

• Used Java Architecture for XML Binding (JAXB) to unmarshalling and marshalling between XSD schema and java objects.

• Used LDAP server provided by Java Decorator API as a means to get JMS connection factories and destinations which are obtained from predefined value in database.

• Used J2EE Standard technologies like JdbcTemplate to connect to both Oracle and DB2 database and set up Data source and JNDI reference in Weblogic Server.

• Developed this whole application from scratch to production with team members and involved in exception handling, code reviews, unit testing and integration testing.

• Coordinated with other teams to call the EBS web-service, create different level log files and set up OVO Monitoring.