Shi D.

Software Engineer

Santa Clara, California, United States16 yrs 1 mo experience

Key Highlights

Expert in AI and High-Performance Computing.
Invented a novel Sparsity Monitor for DNN training.
Led development of GPU-based Distributed Mathematical Library.

Stackforce AI infers this person is a specialist in AI and High-Performance Computing with a focus on deep learning technologies.

Contact

Skills

Core Skills

Machine LearningHigh-performance ComputingDeep LearningSoftware Development

Other Skills

CC++JavaDistributed SystemsAlgorithmsPythonLinuxGPGPUCUDAOpenCLTensorFlowBig DataDeep Neural NetworksBig Data PlatformsAnalytics

About

I am now a Senior Member of Technical Staff at Cerebras Systems. I love research topics related to AI and High Performance Computing. My research interests include Compute Support of Deep Learning, High-Performance Computing, and Big data Platforms/Analytics My GitHub repo https://github.com/shidong-ai My Google Scholar page https://scholar.google.com/citations?user=qtoQQcwAAAAJ&hl=en

Experience

16 yrs 1 mo

Total Experience

Average Tenure

Current Experience

Cerebras systems

2 roles

Senior Member of Technical Staff

Promoted

Aug 2022 – Present · 3 yrs 10 mos

CC++JavaDistributed SystemsMachine LearningSoftware Development+10

MTS, Member of Technical Staff

Sep 2020 – Aug 2022 · 1 yr 11 mos

Amd

Research Intern

May 2018 – Aug 2018 · 3 mos · Santa Clara, CA, USA

Investigate characteristics of data sparsity during DNN training
Led the investigation of the data sparsity patterns and trends during DNN training
Invented a novel Sparsity Monitor for DNN training with extremely low data transfer overhead (pending patent)

DNN trainingData SparsitySparsity MonitorDeep Learning

Samsung electronics

Research Intern

Jun 2015 – Dec 2015 · 6 mos · San Francisco Bay Area

Development of a GPU-based Distributed Mathematical Library (dMath)
Led the development of multiple demo applications of DNN models using dMath
Optimized scalability and memory usage of dMath
Led the development of the batch normalization algorithm on dMath

GPU-based Distributed Mathematical LibraryDNN modelsBatch Normalization AlgorithmDeep Learning

Northeastern university

Research Assistant

Sep 2014 – Aug 2020 · 5 yrs 11 mos · Greater Boston

My research interests at NUCAR include Compute Support of Deep Neural Networks on GPUs, High-Performance Computing, and Big data Platforms/Analytics

Deep Neural NetworksHigh-Performance ComputingBig Data PlatformsAnalyticsDeep Learning

Intel corporation

Software Engineer

Mar 2012 – Aug 2014 · 2 yrs 5 mos · Beijing, China

Development of a physical layer controller for an LTE mobile communication system
Led the design and development of new system features, i.e., downlink & uplink scheduling, downlink HARQ, intra- & inter-frequency measurements, discontinuous reception, and SPS & TTI bundling
Led system performance optimizations such as power savings and robustness
Analyzed, identified and resolved 300+ technical issues during development and verification phases

Physical Layer ControllerLTE Mobile Communication SystemSystem Performance OptimizationsSoftware Development

Mediatek

DSP engineer

Mar 2010 – Mar 2012 · 2 yrs · Beijing, China

Development of a dual-Mode and dual-SIM mobile communication system
Principle developer of new software protocols to support dual-SIM on ARM
Proposed and implemented software solutions to enhance overall performance in channel monitoring, signal strength measurement and power savings
Led system testing, debugging, and troubleshooting

Dual-Mode Mobile Communication SystemSoftware ProtocolsChannel MonitoringSoftware Development