职位描述 1、参与量化交易模型训练框架的开发及性能优化工作; 2、优化CPU/GPU的推理延迟,根据业务编排算子,组织计算流水线,优化算子的执行效率,实现等效可替换算子等工作； 3、提高数据加载速度、缓存利用率,优化分布式训练流程,提升训练中的GPU利用率; 4、设计和实现各类机器学习训练工具,简化从回测到实盘的部署流程,方便研究员管理模型、数据、算法等模块,加速模型的开发和测试 Role Overview: We are seeking a Machine Learning Framework Developer to spearhead the development and enhancement of our quantitative trading model frameworks. This role is instrumental in optimizing our proprietary trading algorithms and streamlining our trading processes. What you'll do: - Collaborate with the quant team to develop and fine-tune the performance of our model training framework. - Optimize CPU/GPU inference latency, architect computational pipelines based on trading strategies, and elevate the efficiency of our algorithmic operators. - Boost data ingestion speeds, maximize cache utility, refine distributed training mechanisms, and ensure optimal GPU resource allocation throughout the model training phase. - Lead the design and implementation of cutting-edge machine learning training tools, ensuring a seamless transition from model backtesting to live trading. You will also be streamlining the workflow for our researchers in managing trading models, data sets, and complex algorithms. 职位要求 1、计算机、数学、统计学或相关专业硕士或博士学位; 2、熟悉C++、CUDA、Python等多种语言,扎实代码功底和实战能力 3、熟练掌握Tensorflow/Pytorch之一,熟悉分布式训练架构与调优 4、熟悉高性能 RDMA 通信等相关知识。 - Master's or PhD in Financial Engineering, Computer Science, Mathematics, Statistics, or a related discipline. - Proficiency in C++, CUDA, and Python with a demonstrable track record of robust coding and practical application in a finance/trading environment. - Comprehensive understanding of Tensorflow or Pytorch, with hands-on experience in distributed training architectures and optimization techniques. - Familiar with high-frequency trading environments, especially with respect to high-performance RDMA protocols.