Dong Ping Zhang, Ph.D. Email & Phone Number

Palo Alto, Ca, Us

Santa Clara, California, Us

Develop and drive cross-organisational technology strategic initiatives. Build a team of senior exceptional technologists covering a range of technical domains.Areas covered: technology innovation and pathfinding in various domains (e.g., ML/AI and data accelerators, ML/AI <-> system architecture, LLM training and inference optimisation, HW & SW co-design, platform QoS, scalable I/O virtualisation, confidential computing), productisation (e.g., tech readiness processes, cross-org engineering & development work and alignment, roadmap, customised solutions partnering with CSPs), and industry standardisation (e.g., OCP, PCI-SIG)

Santa Clara, California, Us

AMD Spotlight Award - 2021, 2022Lead and develop AMD's AI 2030 strategies on graphics, game AI and metaverse.Develop and evaluate business cases for ML/AI technology, long-term strategic direction, market analysis and key growth opportunities.From scratch, build and lead a group of SW engineers working on AI technology innovations in the gaming space, work broadly with executives, architects, business units, ISVs, and academics to generate differentiating technologies and initiatives.Lead AMD's graphics & gaming AI, and inference North Star Initiatives with distributed teams: * Example content creation and runtime app areas: NLP, audio2face, character animation, NPC reinforcement learning, neural rendering, deformable simulation, avatar etc; with Epic UE, Unity etc. * Devise, scope, evaluate technical feasibility, assess resource requirements of projects. * Influence SW engineers and HW architects across multiple functional units to make thorough technical and business assessments of projects and requirements. * Prototype, implement and evaluate software projects, and propose architecture features. * Influence architects to design functionality to better support applications in the long-term product roadmap.Tech lead of ML/AI inference program: * Focused on deep learning based edge inference solutions, SW and HW co-optimisation, architecture feature proposal and evaluation.ML training and inference acceleration in various application domains, e.g, automotive, video conferencing, gaming, graphics.AMD Institutional Research Committee - University RelationTechnical program lead of next-gen GPU arch design - modular graphics: * Topics including chiplet, die stacking, power, thermal, cost estimates, cross-layer co-design (packaging + circuit + arch + app), and software optimisation for architectures in planning. * Tracked progress, guided discussions, highlighted issues, promoted a tight cross-organisational environment to ensure success.

Us

Artificial intelligence + education technology

London, Gb

Proud to serve as the President of this prestigious alumni association for 600+ Imperial College alumni in California. Organise regular social and networking events. Support and promote official college events. Collaborate with college alumni outreach office to provide a network of contacts for our alumni.

Palo Alto, California, Us

Building a general purpose silicon photonic quantum computer1. architecture integration for linear optical quantum computing.2. bridge the gaps between architecture group and the hardware engineering,system and packaging groups.3. hiring, leading and building teams (Linear Optics, Fault Tolerance, Architecture Simulation, Application), defining the structure and scope of the teams and key responsibilities.

San Jose, Ca, Us

Lead the research, development and product effort of building a deep learning semantic search engine, developed and deployed on Google Cloud Platform; hands-on practitioner developing software product using Agile methodology. Areas of interest during this time include: semantic representations of queries and inventories, vector search, natural language understanding, AI dialogues, multi-lingual cross-border trading, online and offline ranking, click prediction and recommendation system.

Santa Clara, California, Us

Technical lead for the advanced memory project of the Exascale computing domain, covering software, runtime, programmability, application and architecture co-design for in-memory computing. Lead and mentor a team of engineers and scientists with responsibility for short and long-term technical planning. Define technical scope and fine-grained work plan, and estimate resource requirements to apply for research grant spanning multiple years. Responsible for project proposal, statement of work, planning and delivery. Filed dozens of patents in the interdisciplinary fields of applications and computer architecture.Lead projects on scaling out deep learning applications on multiple accelerators, including deep belief networks, auto encoder, convolutional neural network, recurrent neural network etc. Design and develop various machine learning approaches (such as graph kernels and deep learning algorithms) for malware detection resulting in two peer-reviewed papers. Develop a data locality and computation management library to adapt to lessons learned from application studies and capabilities developed from the architecture investigation. Aim to provide a clean abstraction to give application developers easy access to multi-node heterogeneous architectures. Lead a project to develop a new bucketized cuckoo hash table design with improved data access efficiency, by reducing the expected cost of positive and negative lookups to fewer buckets maintaining the high load factor. Investigate the scalability of online analytical processing (OLAP) workloads through the evaluation of TPC-H on Ocelot (GPGPU extension to MonetDB); Research on scaling out these applications on in-memory computing systems with multiple nodes. Served on program committee and as reviewer for many international conferences and workshops.

Santa Clara, California, Us

I am leading the advanced memory project in the Exascale compute node architecture and memory technology research domain, including software, runtime, programmability, application and architecture co-design. 1. Creating IPs in the interdisciplinary areas of computer architecture and various application domains including image analysis, data analytics, computer vision and machine learning. 2. Leading a team of engineers, proposing R&D projects, and delivering end products including software implementation and documentations. 3. Implementation and optimization of DNN (deep neural network) algorithms on heterogeneous systems: data and model parallelization on a multi-GPU platform, evaluation of in-memory compute architectures, application of these algorithms in various domains.4. Data analytics: designed and implemented a semantic vector search engine for large scale image clustering and search applications: global- and local-descriptors and classification driven approach followed by verification, with performance and energy evaluations on current and proposed architectural platforms.5. Leading the development of a C++ template library for data-locality aware computation on NUMA-like multiple processor-in-memory architecture.6. Research and development in the domains of online analytical processing with multiple in-memory computing modules and compute migration on NUMA-like architectures. 7. Adapt the HPC applications (CoMD, miniFE, etc) for heterogeneous NUMA node architecture with multiple GPUs, followed by performance and energy optimisation.

Santa Clara, California, Us

My work in AMD Research focuses on:1. Investigation of memory architecture, programmability issues for Exascale systems.2. Co-design of application with simulator API and architecture model. 3. Analysis of computation and memory access patterns of applications, implementation on simulation framework, study of performance and power trade-offs to guide the architecture design.

Santa Clara, California, Us

My role in Heterogeneous System Architecture Group including but not limited to the following: Represent AMD at Khronos Vision Working Group on creating an accelerated computer vision hardware abstract layer that was subsequently released as the OpenVX specification.Develop and map algorithms in large-scale image search, biomedical data analysis, machine learning and natural user interface domains for heterogeneous system architecture. Design and create domain-specific parallel workloads (OpenCL) to validate and refine the system architecture decisions, analyze, profile, optimise the power and performance of these workloads on existing hardware platforms, also perform the projection for future hardware.Proposed and contributed to a hand gesture recognition project and a speech recognition project; Organize Heterogeneous System Architecture (HSA) technical seminars to provide a platform for engineers and scientists to share their technical insights and promote direct information flow and collaboration across organisations.

Santa Clara, California, Us

Heterogeneous System Architecture GroupPropose and lead the development and optimisation of OpenCVCL project (OpenCL accelerated OpenCV routines). This subsequently was released as OCL module, part of the OpenCV-2.4.3: https://opencv.org/opencl/

London, Gb

I worked on designing image analytics and machine learning algorithms to identify and extract biomarkers from brain images to establish the link between brain image changes and cognitive function variation. Collaborated with scientists from Sweden, Finland, Italy and Denmark on the EU PredictAD project.

London, Gb

1. Designed image analysis and machine learning algorithms to automatically classify and segment the anatomical structures affected by Alzheimer’s disease from large volumetric image data.2. Quantitatively analysed a large number of brain images both temporally and across the population to study the development of Alzheimer’s disease, using image segmentation and registration algorithms.

London, Gb

Areas: constrained optimisation, graphics and visualisation, computer vision and virtual reality, data analytics, machine learning, image processing, high performance computing

1. Designed segmentation and visualisation algorithms for small cardiac vascular structures in large-scale data sets.2. Established the collaboration that resulted in a number of publications and conference presentations in the following two years.

Leadership And Strategy

Phd, Computing

Chartered Financial Analyst Level 1

Msc, Advanced Computing