[Platform] Add support for Google vertex AI

[junaidbinfarooq]

Q	A
Bug fix?	no
New feature?	yes
Docs?	yes
Issues	#16 #272
License	MIT

Changes proposed:

• The PR aims to add support for Google's Vertex AI
• Support is added for text generation, embeddings, server tools, etc
• I created a separate directory inside the Bridge directory for the said bridge. Moving forward, we could either create a new namespace named Google and add two namespaces inside it, one each for Gemini and VertexAi, or we could simply keep VertexAi and remove the other, as this api already supports Gemini and more. Either way, Normalize Gemini vs Google #146 would be addressed.
• I exempted the use of api key from the implementation so far as VertexAi supports two auth mechanisms, Api keys and ADC, ie, Application Default Credentials. The former is recommended for testing while the latter is recommended for production, IIRC. However, it does need to install gCloud on the server and locally too (if verifying the behavior without an api key). The Google auth library is also needed for automatic credential discovery.
• More information can be found at https://cloud.google.com/vertex-ai/generative-ai/docs

[faizanakram99]

lgtm, ship it

[junaidbinfarooq]

lgtm, ship it

Docs and AI bundle integration need to be added first.

[OskarStark]

The indention looks wrong

[junaidbinfarooq]

In the PHP doc, the IDE doesn't apparently help with the indentation in this case.
I tried to correct the indentation now.

[chr-hertel]

Thanks @junaidbinfarooq for your work on this - I think we're getting somewhere.
Please see my comments and review once again yourself with those in mind - not sure I catched everything.

In general it great to have that bridge, I still wonder tho about the amount of duplicated code with the Gemini bridge and we would need to support API Keys as well ... but both topics are fine as follow up

[junaidbinfarooq]

Thanks @junaidbinfarooq for your work on this - I think we're getting somewhere. Please see my comments and review once again yourself with those in mind - not sure I catched everything.

In general it great to have that bridge, I still wonder tho about the amount of duplicated code with the Gemini bridge and we would need to support API Keys as well ... but both topics are fine as follow up

@chr-hertel
You're welcome 😄
The duplication is expected as I did not touch the existing Gemini bridge code.
This is what I also pointed out in the PR description about possibly having Vertex AI only as we move forward.
If this is done, there would be no duplication anywhere as far as Google bridges are concerned.
And yes, I do believe this is better done in a follow-up PR after a proper discussion on the topic.

[chr-hertel]

@junaidbinfarooq are all examples that you added working for you?

i need to apply those changes, any clue why?

[junaidbinfarooq]

@junaidbinfarooq are all examples that you added working for you?

i need to apply those changes, any clue why?

@chr-hertel
I just rechecked the vertex ai api and also ran the examples again to verify they are still working as expected. And it is apparent that they are working:

Besides, I didn't use versions in model IDs as that is not recommended, IIRC.

[chr-hertel]

Hmm, super weird 🤔

And where is that $location-aiplatform.googleapi.com/... coming from?
Docs say it's just aiplatform as subdomain:

See https://console.cloud.google.com/vertex-ai/publishers/google/model-garden/gemini-2.5-flash-lite