Agentic AI Reference Architecture

Model Adaptation · Software component

Segment-Routed Reward Scorer

Software componentModel AdaptationModelsarc:SegmentRoutedRewardScorer

A reward scorer that selects among separate reward models trained for different user segments, scoring each response with its segment's model to personalize alignment.

Responsibility. Applies segment-specific preference models instead of a single averaged reward function.

Also known as: Personalized reward models, Per-segment reward models

Variant of Reward Scorer abstract

When to choose. Choose when user segments legitimately hold divergent preferences and sufficient preference data can be collected for each segment.

hostsspecializesis target of alternativeTois target of alternativeToReward Model: hostsReward ModelReward Scorer: specializesReward ScorerPreference Reward Scorer: is target of alternativeToPreference Reward ScorerEnsemble Reward Scorer: is target of alternativeToEnsemble Reward Scorer
Direct neighbourhood (hover for relationship types)

Relationships

hosts structural

alternative to variability

Design guidance

Classification

Risks mitigated
Preference averaging that satisfies no user groupCultural value homogenization

Sources

  1. Ch10.3: T. Nguyen, "RLHF Methodology," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 10.3. ISBN: 9798244538229.