AI / ML
Managed services for training, hosting, and calling machine-learning and generative-AI models — from prebuilt foundation-model APIs to full training and inference platforms. Place it to show where inference or model-serving sits in the system.
Description
The managed AI/ML platforms (AWS SageMaker with Bedrock for foundation models, GCP Vertex AI, Azure Machine Learning with Azure OpenAI Service) all cover the lifecycle from data preparation and training through to hosted inference endpoints, plus API access to prebuilt and generative models. The differences are in emphasis and ecosystem: each bundles its own notebook/pipeline tooling, its own catalogue of first-party and partner foundation models, its own accelerator options (e.g. AWS Trainium/Inferentia, GCP TPUs, GPUs across all three), and its own MLOps and pricing model. To the best of our knowledge these are equivalent for placing model-serving in a topology; choose by the specific models you need, accelerator availability, and how the platform integrates with the rest of your data stack.
Capabilities
- Call prebuilt foundation and generative-AI models via API
- Train and fine-tune custom models on managed infrastructure
- Host models behind scalable, low-latency inference endpoints
- GPU/accelerator-backed compute for training and serving