VT Lab

Job Responsibilities:

- Construct and manage the generative AI model (for footwear and apparel) assets, and the original training dataset;

- Generate fashion design drawings and marketing images with our AI generative models, according to customer requirements;

- Establish an image search engine based on vector databases (e.g., Malvis, FAISS), supporting image searches through descriptions or reference images;

- Build workflows to complete model training such Embedding, LoRA, Hypernetwork, etc., according to requirements;

- Use ComfyUI to construct a workflow pipeline for footwear and apparel design generation, and integrate the pipeline into the APP in collaboration with the system team;

- Optimize APP functions such as partial redrawing, high-definition repair, and composition expansion in Stable Diffusion;

- Build a digital human fashion model asset library for virtual digital models and human models, including face, body, and posture;

- Participate in the research and development of image-based virtual try-on algorithms, and test various algorithms in different scenarios as required.

Position Requirements:

- Master's degree in Computer Science, Mathematics, or related fields, with a foundation in computer vision and deep learning;

- Familiarity with the SOTA models for various common tasks in the field, capable of model selection and implementation;

- Familiarity with diffusion generative models, with a deep understanding of the parameters and extension functions of Stable Diffusion;

- Proficient in training various generative image models and able to independently complete the training and tuning of checkpoint models.

- Possess good engineering R&D capabilities, strong practical skills, fluency in English, and following the development of forefront technology.