Demystifying Usage of AI and ML with Azure Serverless

It was an honour for us to deliver a session at the 2019 Azure AI tour in Mumbai. We were invited to hold a session on “Artificial Intelligence / Machine Learning with Azure Serverless”.

The conference was held at the Microsoft office, Santacruz East, Mumbai. This was our first time attending the Microsoft Azure conference. It was great fun, and the conference was very well organized.

It had around 70 attendees from different states and 7 speakers delivered 7 sessions.

The event kicked off on a bright note. Noelle LaCharite, the Developer Experience Lead for Applied AI at Microsoft covered various aspects of AI, with ease of learning and provided code base for developers. She also presented a handful of demos related to a few cognitive services.

Gandhali Samant (Sr. S/W Engineering Manager – Financial Services Cloud Architect at Microsoft), presented different business case studies where Microsoft Azure AI was widely and successfully being used.

Along with the informative presentation, she also presented a few videos documented as a part of AI implementation.
Our representative had the pleasure to deliver a 45-minute session at the conference. It was wonderful interacting with a lot of Azure architects, .Net Devs and data science experts.

Here are the quick highlights of our session:

The session started with Azure AI computer vision service, custom vision service, and Azure Functions.

Furthermore, we demonstrated the usage of services in Azure functions by bindings and Azure Signalr. We also spoke about code in .Net and Python.

Some of the highlights of our session are as follows:

Fig 1: Serverless Architecture: The What & Why

We further explained solution implementation using Azure function .net, Python SDK and Azure SignalR.

• Azure function

  Triggers Timer, Http, Blob, Cosmos Db, Queue, Event Hub

• Bindings

  Blob and SignalR

One of the most pivotal things we displayed, were the use cases for Computer Vision Service. It customises your own state-of-the-art computer vision models for your unique use case.

You just need to upload a few labelled images and let Custom Vision Service do the hard work. With just one click, you can export trained models to be run on device or as Docker containers.

The use cases included the following:

Use case 1:

Requirement:

When an image is uploaded:

Fig 2: Use case: When an image is uploaded

Use case 2:

Requirement:

Predict an anomaly:

Fig 3: Use case: Predict an anomaly

Our overall objective was to let people know that serverless computing is a relatively new paradigm in server technology. It will help organizations to convert large functionalities into smaller, discrete on-demand functions.
These can be invoked and executed through automated triggers and scheduled jobs.

Additionally, we would like to mention that we enjoyed the entire event as it shared valuable information on AI, cognitive services, IoT, Serverless, and cloud concepts.

The enthusiasm among all the attendees was commendable. Their utmost excitement to learn about artificial intelligence was evident!

We must say we are grateful to Microsoft and Azure India for this wonderful opportunity. Let us learn and grow together!

Read our blog titled ‘Introduction to Serverless: Simplifying Modern Application Development’

Write to us with your views related to Serverless and artificial intelligence. Visit us at Nitor Infotech to learn about our leading-edge software services.