北京招聘

Job Description:

· Develop the neural network compiler for AI processor.

· Develop the hardware emulator for AI processor.

· Work closely with hardware engineers and algorithm engineers to improve performance of AI processor.

Qualification:

· Bachelor's degree in Computer Science.

· 3+ years’ experience in software development.

· Strong C/C++, Python programming skills and unix/linux experience is required.

· Familiar with computer vision, deep learning and optimization methods, is a plus.

· Experience on DL frameworks, such as TensorFlow/Pytorch/Caffe, is a plus.

· Experience on neural network compiler development, such as XLA and TVM, is a big plus.

· Good English reading and writing skills.

工作描述：

· 为AI处理器开发编译器

· 为AI处理器开发模拟器

· 与硬件和算法团队协同工作优化并提高AI处理器运行性能

任职要求：

· 本科以上，计算机相关专业

· 3年以上软件开发经验

· 精通c/c++, python 编程语言，能够在linux下开发

· 熟悉计算机视觉，深度学习者优先

· 使用过Tensorflow/Pytorch/Caffe，等深度学习框架者优先

· 有开源编译器开发经验者优先，例如XLA, TVM

· 良好的英语读写能力

作为业界领先的集成电路设计公司之一，澜起科技致力于为云计算和人工智能领域提供高性能芯片解决方案。公司在内存接口芯片市场深耕十余年，先后推出了DDR2、DDR3、DDR4系列高速、大容量内存缓冲解决方案，以满足云计算数据中心对数据速率和容量日益增长的需求。澜起科技发明的DDR4全缓冲“1+9”架构被JEDEC采纳为国际标准，其相关产品已成功进入全球主流内存、服务器和云计算领域，占据国际市场的主要份额。

2016年以来，澜起科技与英特尔及清华大学鼎力合作，研发出津逮系列CPU。基于津逮CPU及澜起科技的安全内存模组而搭建的津逮服务器平台，实现了芯片级实时安全监控功能，为云计算数据中心提供更为安全、可靠的运算平台。此平台还融合了先进的异构计算与互联技术，可为大数据及人工智能时代的各种应用提供强大的综合数据处理及计算力支撑。

澜起科技成立于2004年，总部设在上海并在昆山、中国澳门、美国硅谷和韩国首尔设有分支机构。

Montage Technology, one of the leading fabless semiconductor company, is committed to providing high-performance IC based solutions for cloud computing and artificial intelligent (AI) markets. Dedicated for more than 10 years in the memory interface technology, the company is able to deliver high-speed, large-capacity memory buffer solutions of DDR2 to DDR4 to meet the growing demand in data centers. Montage’s invention of DDR4 fully buffered “1+9” distributed architecture is adopted by JEDEC as an international standard, and its related products have successfully entered the industry mainstream memory, server and cloud computing fields and served the majority of the global market.

Since 2016, Montage has been cooperating with Intel and Tsinghua University to develop Jintide CPU. A high-performance Jintide server platform combining Jintide CPU with Montage's hybrid security memory module (HSDIMM) has been launched. The platform has realized real-time security monitoring function at silicon level and can play an important role in the information security field. This architecture also incorporates advanced heterogeneous processing and interconnect technologies for future AI and big data applications, which provides strong support of comprehensive data processing and computing power for various applications in the AI era.

Established in 2004, Montage Technology is headquartered in Shanghai and has branches in Kunshan, Macao, Silicon Valley of USA and Seoul of South Korea.