<p>Duties (including internship and possible permanent position)</p>
<p>1. For embodied intelligent large models, we will conduct data collection, crawling, cleaning, processing, and optimization through the whole process, and provide high-quality data for model pre-training and fine-tuning. Finally, we will form a data-model closed loop.<br>2. Build a data-centric platform and tools to unify the collection, processing, querying, and management of multimodal data.<br>3. Use CV algorithms, multimodal models, etc. to process video, image, etc. multimodal data, such as object detection, video description, etc.<br>4. Guide and manage large-scale data annotation work, and be able to guide model training from the perspective of data.</p>
<p>Qualifications:</p>
<p>1. Have good model algorithm construction ability, be able to independently train CV models or multimodal models based on open source projects. Those who have in-depth and comprehensive understanding of natural language processing (NLP), computer vision (CV) and multimodal models are preferred.<br>2. Have experience in model inference optimization acceleration, familiar with quantization, pruning, etc., and able to meet the large-scale inference needs of business requirements. Familiar with VLLM and other inference acceleration frameworks.<br><br>4. Familiar with Python, git, Linux system, solid data structures and algorithm foundation, good programming foundation.<br>5. Familiar with computer networks and computer organization principles, with solid foundations in data structures and algorithms.<br>6. Familiar with Hadoop, Spark, Flink, etc., big data processing frameworks, with actual project experience. <br><br>1. Have the ability to quickly learn and master new field knowledge, as well as good team spirit<br>2. Bachelor's degree or above in Computer Science, Software Engineering, or other related fields.</p>