Job Responsibilities:
1. Participate in the R&D of large models such as NLP, multimodal, and CV, be responsible for the capability construction in one or more areas including NLP, natural images, documents, remote sensing, video, 3D, etc., undertake horizontal technical direction research, including encoders, modal alignment, training methods, contextual learning, object hallucination, search enhancement, Agents, etc.
2. Support the construction of relevant L1 layer models, participate in completing customer communication, requirement collection, technical analysis, and other work, and support the achievement of L1 layer effects and the final project landing through the construction of L0 layer capabilities.
3. Participate in the technical planning of NLP & multimodal large models, undertake insight analysis, innovative exploration, university cooperation, technology introduction, etc., in specific technical fields, and be responsible for the technical competitiveness in the relevant fields.
Job Requirements:
1. Technical Vision: Have an in-depth understanding of the current status and development trends of large models such as NLP, multimodal, and CV, have a clear and specific understanding of the current status of large model technologies of leading manufacturers, be familiar with the full-stack technical system of NLP & multimodal large models including model architecture, training methods, evaluation methods, data construction methods, etc., and have in-depth insights into key issues in multiple technologies.
2. Technical Accumulation: Have profound technical accumulation in one or more fields such as NLP, CV, OCR, remote sensing, 3D, AIGC, including traditional classification, detection, segmentation, visual language pre-training, and end-to-end modeling, etc., be proficient in model pre-training and fine-tuning methods in the field, master the method of effect problem location analysis, be proficient in the theory of basic network architectures such as CNN, Transformer, and have experience in model architecture optimization. Preferred conditions: high-quality paper publications in related fields, experience in NLP.
3. Successful Experience: Have participated in the R&D of mainline models of main products in the company, undertaken key roles such as model pre-research, model planning, model development leadership, etc., and played a leading role in one or more aspects such as model technology landing, model effect improvement, model problem solving. Candidates leaning towards technical management need to have solid cases of guiding technical teams to support business success.