AI Model Policy Trainer – Image Evaluation

• Evaluate generated images against prompts for adherence, composition, realism, style consistency, and technical defects
• Compare images side by side and select the stronger image with an evidence-based rationale
• Classify images under customer policies covering sexual content, violence, hate symbols, real-person likeness, intellectual property, and depictions of minors
• Select defensible classifications for ambiguous images and write concise rationales citing rubric language and visual details
• Separate prompt failure, policy violation, and personal preference judgments
• Write and refine prompts to test model quality and safety boundaries
• Identify rubric gaps, contradictions, and emerging edge cases and raise them with project leads and policy teams
• Participate in calibration discussions and update judgments based on stronger reasoning
• Apply customer policy consistently without substituting personal taste or beliefs
• Maintain accuracy and consistency across hundreds of visually similar evaluations
• Engage regularly with sensitive imagery within structured evaluation frameworks and professional guidelines

• Deep visual judgment from photography, illustration, concept art, art direction, retouching, photo editing, VFX, visual design, generative image tools, or visual content moderation
• Ability to evaluate image adherence, composition, realism, style consistency, and technical defects
• Familiarity with generative image tools such as Midjourney, Stable Diffusion, ComfyUI, Flux, DALL-E, and Ideogram
• Knowledge of generative image model failure modes, prompt quirks, and safety-filter bypasses
• Experience moderating or reviewing visual content at scale and applying policy taxonomies under time pressure
• Ability to identify subtle visual details such as extra fingers, mismatched shadows, brand marks, and recognizable faces
• Ability to maintain rubric consistency across hundreds of similar evaluations
• Ability to explain judgment calls clearly and precisely in writing
• Ability to distinguish personal taste from customer standards
• Maturity and sound judgment when handling sensitive imagery and difficult subject matter
• Ability to engage with sensitive material carefully, responsibly, and sustainably
• A degree and technical background are not required
• Prior AI evaluation experience is helpful but not required
• Must be authorized to work lawfully in the United States for Handshake
• Must work 8AM–5PM PT, Monday through Friday

Demonstrates deep visual judgment and the ability to evaluate image adherence, composition, realism, and technical defects while applying customer policies consistently. Engages with sensitive imagery responsibly and maintains rubric consistency across evaluations.

• Visual Judgment
• Generative Image Tools
• Content Moderation Experience
• Policy Application
• Attention to Detail

• Image Evaluation
• Composition Analysis
• Technical Defect Identification
• Prompt Writing
• Policy Classification

• Clear Communication
• Sound Judgment
• Maturity

• Visual Content Moderation
• Generative Image Models
• Policy Taxonomies
• Sensitive Imagery Handling

• Midjourney
• Stable Diffusion
• ComfyUI
• Flux
• DALL-E
• Ideogram