Feature Overview (mandatory - Complete while in New status)

• SDG Hub v2 is a major design change that will improve UX for building new data generation flows.

Pain Points with v0 Architecture:

• YAML Based Definition: The current design is tightly coupled with the YAML based definition of flows. This makes it difficult to extend the framework to support pythonic definition.
• No explicit connections between blocks: There's no explicit way to connect blocks together. The only way to achieve this is to wrap them in a pipeline
  Sequential Execution: The current pipeline object is basically a list of blocks, which triggers the sequential execution of blocks. This makes it difficult to create more complex workflows with parallel processing, conditional branches, or feedback loops
• Tight Coupling: Blocks in the current design are tightly coupled with their execution context. Executing a couple of blocks require them to be wrapped in a pipeline - which needs a flow yaml for definition - which needs registration of llm prompts.
• This arised due to the previous (now stale) requirement of yaml driven definition.

Goals (mandatory - Complete while in New status)

• Allow users to build data generation workflows through Pydantic flows. 

Requirements (mandatory -_ Complete while in Refinement status):

• Example Notebook with new usage patterns for prompt tuning 
• Example Notebook with new usage patterns for existing pipelines - knowledge, skills, reasoning, annotation
• Example Notebook with new usage patterns for complex workflows

Done - Acceptance Criteria (mandatory - Complete while in Refinement status):

• Validated example notebooks compatible with vllm versions 
• Notebooks tested against RHAIIS, RHEL AI, RHOAI