AI ML with Azure Serveless
- 2. A D https://www.linkedin.com/in/akshaydeshmukhis Speaker @Dear Azure Blogger, MVP @C Sharp Corner https://www.c-sharpcorner.com/members/coder30 Author @Dot Net Tricks https://www.dotnettricks.com/mentors/akshay-deshmukh Love to use Azure ML, IoT, Microsoft Bot Framework, .Net Core, Angular Love to code in C#, Python, Scala, TS, JS
- 3. Serverless - what why Azure Functions – what HoW • usecase 1 – .NET SDK + computer vision API • usecase 2 – Python SDK + scikit-learn Agenda
- 4. Serverless - what why At the end of the day, you need servers to deliver software server management, capacity planning Software development approach – • managed compute service (FaaS) • Leverage SaaS products Reduced Ops, Focus on Business Logic, Reduced Time to Market Serverless doesn’t mean NO SERVERS
- 5. Azure Functions – what • to run a script or piece of code in response to event • to run code on-demand • Pay only for the time your code runs • scale as needed – by Azure • triggers bindings Timer, Http, Blob, Cosmos Db, Queue, Event Hub Many more…. • Plans – Consumption, App Service • Dependency management - supports NuGet and NPM
- 6. Azure Functions – how(use case 1) Requirement: When image is uploaded • identify applicable tags • identify face, gender, age • suggest possible description • send real-time notification
- 7. Azure Functions – how(use case 2) Requirement: Predict anomaly • trained using scikit-learn • saved using pickle on Azure blob • solution to be available as service Solution:
- 8. Summary: • Triggers and binding – connect with other services • Stateless and async – enables scalable AI/ML solution • Supported languages – C#, F#, Java, Python many more….