Skip to content
All jobs & gigs

Video Model Evaluation Expert (Audio & Video Generation) (视频模型评测专家(音视频生成方向))

ShengShu Technology (Vidu) · Beijing, China

Full-timePosted Nov 18, 2026

ShengShu Technology (Vidu) is hiring: Video Model Evaluation Expert (Audio & Video Generation) (视频模型评测专家(音视频生成方向)) (Beijing, China). ShengShu Technology is the Beijing-based company behind Vidu, an AI video-generation model and creation platform. Main focus: Evaluation system construction: Build and continuously iterate the effect evaluation system of the audio and video generation model, including scoring dimensions and scoring rules, test case library, prompt word set and evaluation data set to ensure that the evaluation conclusions are objective and reproducible.

Key responsibilities

  • Evaluation system construction: Build and continuously iterate the effect evaluation system of the audio and video generation model, including scoring dimensions and scoring rules, test case library, prompt word set and evaluation data set to ensure that the evaluation conclusions are objective and reproducible;
  • Industry scene evaluation: For advertising, e-commerce, film and television, short plays, short videos and other landing scenes, evaluate the model's ability to follow instructions, movement quality, subject and scene consistency, physical rationality, composition and lens language, light and shadow color, emotional expression, audio and video synchronization, etc.;
  • Boundary use cases and attribution: Design boundary use cases such as complex motion, multi-person interaction, camera movement, long-term consistency, etc., and establish and maintain a problem attribution label system;
  • Comparative analysis of competing products: Continuously track the performance of competing product models at home and abroad, conduct comparative testing, and analyze the strengths, weaknesses, and sources of gaps of each company;
  • Evaluation process management: organize scoring calibration and consistency sampling, manage the output quality of the outsourced evaluation team, and control the accuracy of evaluation conclusions;
  • Closed-loop output: Output clear, objective, and reproducible evaluation reports, feed badcase and attribution conclusions back to the algorithm and product team, and drive model iteration and product decision-making.

Requirements / qualifications

  • Bachelor degree or above, majoring in film and television production, directing, photography, editing, digital media, animation, computer and other related majors are preferred;
  • Have experience in film and television comics, advertising, e-commerce, short videos or pan-entertainment industries, have in-depth experience in mainstream video generation tools, and have a practical understanding of the effects and controllability differences of different models;
  • Have good film and television aesthetics and professional judgment, be able to accurately identify picture, movement, timing and narrative logic issues in video generation, and be able to break down subjective feelings into quantifiable evaluation criteria;
  • Clear logic and smooth communication, able to independently complete test case design, evaluation specification writing and analysis report;
  • Work objectively and rigorously, be responsible for evaluation conclusions, and be able to adapt to the high-density evaluation rhythm under rapid model iteration.

Nice to have

  • Bonus points
  • Have experience in large-scale model evaluation, competitive product analysis, and experience in building an evaluation system from 0 to 1;
  • Master Python, SQL, and Excel, and be able to write automated evaluation scripts or do data analysis and visualization;
  • Understand mainstream evaluation benchmarks such as VBench and subjective evaluation methodologies such as MOS and Arena;
  • Have editing, photography, storyboarding, animation or visual design capabilities, and can attach a portfolio;
  • Have the ability to read English materials and be able to follow up on overseas models and industry trends.

Team: R&D Location: Beijing, China Employment type: Full-time Posted date: 2026-08-28 Original title: 视频模型评测专家(音视频生成方向) Original posting is in Chinese; description translated to English.