Roles & Responsibilities
1 .Responsible for developing the algorithm application and platform implementation of machine learning models, and meeting the business personalized model training and deployment needs, including implementing and improving model fine-tuning and business adaptation according to business needs
2. Responsible for business scenarios Language model training optimization, inference acceleration, deployment optimization, etc.
3. Implement knowledge enhancement and the application of semi-parametric model technology, and optimize model effects to improve the matching of business requirements
4. Optimization Model fine-tuning algorithms and engineering implementation, etc.
5. Implement model GPU inference deployment optimization, model distillation, etc.
6. Participate in cutting-edge research on technologies related to NLP, knowledge graphs, and multi-modal model implementation, Drive technology development and implementation.
Qualifications
1. Master's degree or above in computer, electronics, mathematics, machine learning or related majors
2. Have at least one of the following experiences: distributed training of language models And inference engineering implementation, large-scale high-concurrency backend system development, in-depth application of knowledge graph and natural language processing. It is required to be proficient in Python, familiar with commonly used data structures and algorithms, proficient in at least one language among C/C++/Java/Scala/Go, and familiar with mainstream machine learning frameworks. Have excellent and agile engineering capabilities, logical thinking and communication skills, strong enthusiasm and self-motivation for AI, keep up with the progress of the AI open source community and have done hands-on practice.
3. Bonus points: Proficient in practicing distributed training framework Complete business-oriented model development and optimization experience As the first author, he has published top international conference papers in the fields of language model, knowledge graph, reinforcement learning, etc.