What this PR does / why we need it:

Adds comprehensive blog post documenting Feast's unified transformation framework, covering the Transformation API, multiple compute engines, and migration patterns.

Changes

Blog Post (infra/website/docs/blog/transformation-framework.md)

• Documents unified Transformation API with 6 execution modes (Python, Pandas, Spark, Spark SQL, Ray, SQL, Substrait)
• Explains ComputeEngine architecture (LocalComputeEngine, SparkComputeEngine) with DAG-based execution
• Includes mermaid diagrams for architecture and feature build flow
• Provides migration guide from legacy On Demand Feature Views
• Contains real-world use cases and code examples
• Authors: @HaoXuAI and GitHub Copilot

Documentation Updates (docs/getting-started/architecture/feature-transformation.md)

• Added callout linking to blog post
• Documented available TransformationMode options with use case guidance

Example Usage

from feast.transformation.base import transformation
from feast.transformation.mode import TransformationMode

# Define transformation with decorator
@transformation(mode=TransformationMode.SPARK)
def transform_features(df: DataFrame) -> DataFrame:
    return df.withColumn("scaled", df["feature"] * 2)

# Apply to FeatureView
feature_view = FeatureView(
    name="driver_stats",
    feature_transformation=transform_features,
    schema=[...],
    source=data_source,
)

Which issue(s) this PR fixes:

Addresses documentation request for transformation framework work from PRs #5181, #5223, #5278, #5185, #5281, #5287, #5354, #5445, #5521, #5523, #5666, #5611, #5612, #5502, #5650, #5629

Misc

Blog post covers architectural improvements that unify feature transformation logic across compute engines while maintaining backward compatibility with existing APIs.

• Fixes Add blog post on new transformation framework #5856

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.