AI Generative Art Specialist
An AI Generative Art Specialist bridges creative vision with technical AI tooling to produce novel visual content, transforming pr…
Skill Guide
The systematic process of designing, testing, and refining input instructions to elicit optimal, predictable, and contextually relevant outputs from large language models.
Scenario
Build a system to draft consistent customer support emails based on short issue descriptions.
Scenario
Create a prompt that transforms a rough feature idea into a structured PRD with user stories, acceptance criteria, and non-functional requirements.
Scenario
Design a system where one LLM acts as a 'Manager' to delegate research, drafting, and editing tasks to specialized 'Worker' LLMs, all via prompts.
CO-STAR structures complex instructions; CoT elicits step-by-step reasoning for logic problems; Few-Shot provides in-context examples for format/tone; Decomposition breaks monolithic tasks into manageable prompts.
Arena compares model outputs; Promptfoo enables automated, test-driven prompt development against custom rubrics; Preference datasets provide benchmarks for aligning with human judgment.
Answer Strategy
Use the CO-STAR framework to structure the answer, emphasizing audience (Objective) and style/tone differentiation. Sample Answer: 'I would create two prompt variants using CO-STAR. For executives, the Objective is strategic impact, Style is concise and business-focused, Audience is non-technical. For engineers, the Objective is technical specification, Style is detailed and precise, Audience assumes technical depth. The core Context (the document) is the same, but constraints and output format differ significantly, requiring separate prompt templates tested on past documents.'
Answer Strategy
Tests debugging methodology and understanding of failure modes. Sample Answer: 'When a model hallucinated API endpoints, I applied a systematic diagnosis: 1) Checked for prompt ambiguity (the request was underspecified), 2) Added a constraint to 'cite only from the provided text,' 3) Shifted from a creative (high temperature) to a deterministic (low temperature) setting, 4) Introduced a few-shot example with correct citation format. The root cause was a combination of vague instruction and high randomness. I now always include source-anchoring constraints for factual tasks.'
1 career found
Try a different search term.