Ethan Park

Tunes role persona communication style for specific audiences and formality levels.

Calculates confidence scores for self-consistency answers based on agreement patterns and reasoning quality.

Plans parallel tree-of-thought exploration with resource allocation and branch synchronization.

Limits LLM output vocabulary to specific allowed terms with synonym mapping and violation handling.

Adapts existing prompts to new domains while preserving structural patterns and intent.

Defines when reflection loops should terminate based on quality convergence and diminishing returns.

Identifies atomic task units that cannot be further decomposed while maintaining independence.

Defines clear expertise boundaries for role personas with appropriate uncertainty and deferral behaviors.

Tunes chain-of-thought reasoning verbosity to match audience needs and token budgets.

Balances few-shot example sets with appropriate positive and negative examples for clear boundaries.

Disambiguates user intent from ambiguous inputs using context signals and pattern matching.

Designs structured markdown output formats with appropriate hierarchy, formatting, and navigation aids.