Prompt Optimization Engineer

Responsibilities 1. Possess a full-chain understanding ability from topic selection, content form to final presentation effect, covering various content forms such as copywriting, images, videos, etc., not limited to writing the Prompt itself. 2. Sort out the actual production scenarios of editing, operations, AI generation, and editing, and define reusable text-based Prompt specifications (titles, copy, SEO content) and visual Prompt specifications (covers, image generation), establishing structured templates and a negative prompt word library. 3. Conduct systematic comparative testing on mainstream text large models, image and video generation models, recording the model capability boundaries, parameter effects, failure modes, and cost differences in different content scenarios, being able to make independent judgments in one's main direction and provide fallback. 4. Design A/B tests and quantitative evaluation indicators to measure usability, consistency, click-through rate/aesthetic quality, generation time, and cost per material, driving optimization with data rather than subjective feelings. 5. Collaborate with operations, AI generators, and development teams to consolidate validated Prompts and evaluation methods into platforms, workflows, or automated evaluation tools, allowing good experiences to be reused and promoted. 6. Possess strong self-drive and review awareness, continuously track model updates and industry practices, systematically organize records of pitfalls, failure modes, and best practices as team assets, and organize internal training to promote the formation of unified prompt engineering standards within the team. Requirements 1. Bachelor’s degree or above, preferably in Chinese, journalism, design, computer science, artificial intelligence, or related fields. 2. More than 2 years of experience in AIGC, Prompt engineering, content production, or model evaluation, with verifiable project cases. 3. Familiar with mainstream text large models and text-to-image, video generation models, understanding different methods such as copy generation, reference images, keyframes, shot control, style transfer, and consistency control. 4. Strong written expression, structured thinking, and experimental design abilities, able to transform personal experiences into reusable templates, processes, and evaluation standards. 5. Able to read model documentation, English materials, and community cases, independently and quickly verify new technologies and draw conclusions. 6. Recognize remote collaboration methods, able to be online and responsive during core working hours, and adapt to high-frequency communication and rapid iteration pace. 7. Sensitive to content quality and data results, with cross-team communication and training output capabilities. Bonus Points 1. Proficient in Python scripting, data processing, or automated evaluation capabilities. Familiar with ComfyUI, LoRA, model fine-tuning, or local inference.