Talk: Making the Internet polite, and the architecture behind
Speakers directory
Speaker:
Ivan Culjak
Talk description
Title:
Making the Internet polite, and the architecture behind
Short synopsis:
Most serverless examples resemble Hello World samples, but what happens when you’re entrusted with a task of building a scalable, and resilient SaaS? Or even “worse” a SaaS around an existing library or service? Come and see a step by step guide with explanation directly from cloud trenches.
Max size: 500 chars
Long synopsis (optional):
Whenever you allow your users to send any type of content, you're taking a risk that that content will either offend someone or at least be contrary to your rules & policies. We're mostly solving those types of problems by combining plain ignorance, automatic verifications against a dictionary of some sort, and a bunch of students who have the "honor" to go through the various content whole day long. That's where Azure Cognitive Services come to the rescue (naturally, there are other vendors) by allowing us to easily, and automatically analyze text, photos, and videos and all of that in multiple languages. Naturally, the decision whether some content is passable or not is to be left to the rules & policies of each system, but the potential of getting the info about someone threating someone else, or someone being in a big depression is mesmerizing. It's not a huge deal to implement those Azure services into your own system, but not everyone has time, and expertise to design, implement, and maintain a proper architecture that will keep the system running across regions, without ever blocking, nor crashing the primary system. Because of that, when I've started pilot projects for a huge TV station, and a "social network" for kids the decision was to build a "wrapper" around Azures services, and instead of integrating everything into each system, I've actually built a "plug&play" SaaS/API that can be easily integrated. This talk is about the architecture of the beforementioned SaaS/API. The ingredients for the recipe are a bit of Cognitive Services, a bit of containers, a bit of Functions (Durable and "normal"), a bit of Event Grid, some ServiceBus, some storage, and DB, combined with a tad of API Management and ML. Come and see how to mix all those ingredients properly ;)
Max size: 5000 chars
Tags:
Speaker directory:
Listed in directory
Not listed
Speakers directory