Create How_Conxai_uses_Knative_Eventing_to_provide_AI_APIs_in_the_con… #6167
Conversation
…struction_industry.md
|
Welcome @tikr7! It looks like this is your first PR to knative/docs 🎉 |
knative-prow
bot
added
the
size/S
✅ Deploy Preview for knative ready!Built without sensitive environment variables
To edit notification comments on pull requests, go to your Netlify site configuration. |
|
@tikr7 thanks a lot! Can you please add your blog article to the navigation? Done in here -> https://github.com/knative/docs/blob/main/blog/config/nav.yml#L51 |
| @@ -0,0 +1,24 @@ | |||
| ## How Conxai uses Knative Eventing to provide AI APIs in the construction industry | |||
|
|
|||
There was a problem hiding this comment.
I'd add an author, like
Author: NAME, TITLE @ SHOP
…struction_industry.md
Done. If something is wrong, pls do let me know. |
|
|
||
| First, the customer uploads an image to be blurred via our API, which saves the image on S3. The S3 event is routed to our AWS EventBridge. The TriggerMesh AWSEventBridgeSource automatically configures an Amazon SQS queue and EventBridge Rule to forward the S3 event into that Amazon SQS queue. Because TriggerMesh has been deprecated, we will switch to Knative Serving [IntegrationSource](https://knative.dev/blog/articles/consuming_sqs_data_with_integrationsource/) approach with AWS SQS in the future. Then it passes that event into the Knative Broker which is backed by Amazon MSK - Kafka. Since we heavily rely on S3 key/file names, we wrote our own router service in Golang which modifies the EventType using regex patterns and sends it back to the Knative Broker. | ||
|
|
||
| Another Knative Trigger moves it into our transformer which handles preprocessing first. As the next step, the transformer calls the predictor (Nvidia Triton) directly over HTTP. We use the "Collocate transformer and predictor in same pod" [feature](https://kserve.github.io/website/0.13/modelserving/v1beta1/transformer/collocation/) to maximize inference speed and throughput. After receiving model results, the transformer does the postprocessing and finally saves the model results on S3. |
There was a problem hiding this comment.
Could we have a little sub-header above this blog as well?
Saying something about KServe / AI ?
There was a problem hiding this comment.
Thank you for the feedback. I integrated something about Kserve.
There was a problem hiding this comment.
Thanks for sharing your adoption story @tikr7 !
In the future, I would love to know more about your router server to see if Knative could provide a generalized version of it and, at the same time, remove that from your plate.
A very small comment/suggestion
|
|
||
| ## Our blurring product step by step | ||
|
|
||
| First, the customer uploads an image to be blurred via our API, which saves the image on S3. The S3 event is routed to our AWS EventBridge. The TriggerMesh AWSEventBridgeSource automatically configures an Amazon SQS queue and EventBridge Rule to forward the S3 event into that Amazon SQS queue. Because TriggerMesh has been deprecated, we will switch to Knative Serving [IntegrationSource](https://knative.dev/blog/articles/consuming_sqs_data_with_integrationsource/) approach with AWS SQS in the future. Then it passes that event into the Knative Broker which is backed by Amazon MSK - Kafka. Since we heavily rely on S3 key/file names, we wrote our own router service in Golang which modifies the EventType using regex patterns and sends it back to the Knative Broker. |
There was a problem hiding this comment.
| First, the customer uploads an image to be blurred via our API, which saves the image on S3. The S3 event is routed to our AWS EventBridge. The TriggerMesh AWSEventBridgeSource automatically configures an Amazon SQS queue and EventBridge Rule to forward the S3 event into that Amazon SQS queue. Because TriggerMesh has been deprecated, we will switch to Knative Serving [IntegrationSource](https://knative.dev/blog/articles/consuming_sqs_data_with_integrationsource/) approach with AWS SQS in the future. Then it passes that event into the Knative Broker which is backed by Amazon MSK - Kafka. Since we heavily rely on S3 key/file names, we wrote our own router service in Golang which modifies the EventType using regex patterns and sends it back to the Knative Broker. | |
| First, the customer uploads an image to be blurred via our API, which saves the image on S3. The S3 event is routed to our AWS EventBridge. The TriggerMesh AWSEventBridgeSource automatically configures an Amazon SQS queue and EventBridge Rule to forward the S3 event into that Amazon SQS queue. Because TriggerMesh has been deprecated, we will switch to Knative Eventing [IntegrationSource](https://knative.dev/blog/articles/consuming_sqs_data_with_integrationsource/) approach with AWS SQS in the future. Then it passes that event into the Knative Broker which is backed by Amazon MSK - Kafka. Since we heavily rely on S3 key/file names, we wrote our own router service in Golang which modifies the EventType using regex patterns and sends it back to the Knative Broker. |
There was a problem hiding this comment.
@pierDipi nice little detail, thank you! I changed "Serving" to "Eventing" :)
…struction_industry.md
knative-prow
bot
added
size/M
…struction_industry.md
…struction_industry.md
…struction_industry.md
We can hop on a call at some point. Let's sync on Slack. I texted you a message. |
|
/approve If anyone has further edits feel free to do it in a follow up. |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: dprotaso, matzew, tikr7 The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
knative-prow
bot
added
the
approved
@matzew as you requested I wrote a small blog post how we use Knative Eventing =)