1. Use AWS ECS + Spot Instance to reduce about 30% server cost 2. Use Terraform to deploy AWS server automatically as implementing Infrastructure as Code, and make it synchronized between staging and production environments. 3. Develop backend API with GraphQL to make frontend easier and more intuitive to use API, and less query to request in the same time. 4. Using Rspect to implement unit test, and deploy to AWS server with AWS Codepipeline with overall 98% unit test coverage. 5. Deploy to server in pure containerized environment with AWS ECS + EC2, all infrastructure are in VPC environment with ALB as endpoint to communicate with client, and implement rolling deployment by Codepipeline to ECS to make it zero downtime. 6. Use GitLab CI to deploy frontend automatically to AWS S3 + CloudFront CDN 7. Use AWS Lambda with Ruby/Node.js to develop chatbot and crawler, and save crawling result into AWS DynamoDB 8. Develop streaming server: use ffmpeg and Nginx-rtmp to develop real time streaming server

1. Install server and manage bare metal machines, including purchasing decision. 2. Image Recognition: Using OpenCV to develop manufacturing automation and deploy the code into embed machine