Staff Software Engineer, XR ML Optimization

Samsung Research America (SRA); Mountain View (CA); On-site; USD 130,000 - 180,000; Full time

Job summary

Samsung Research America is seeking an XR ML Optimization Engineer to join the MPS team in Mountain View, CA. The role focuses on optimizing software performance for XR technologies, requiring extensive experience in hardware/software co-design and deep learning. Candidates should possess strong analytical skills and a passion for innovation in the XR field.

Qualifications

• 8+ years of experience in software/hardware co-design & optimization.
• Deep understanding of CPU, GPU, DSP architectures.
• Familiarity with ML libraries & frameworks like PyTorch, Tensorflow.

Responsibilities

• Analyze XR workloads to identify bottlenecks in hardware and software.
• Propose new HW/SW co-optimization methodologies.
• Build simulations and performance models for decision making.

Skills

Software/hardware co-design

Performance optimization

Deep Learning

C/C++

Python

Teamwork

Communication

Education

Bachelor’s or Master’s or PhD in Computer Science/Engineering

Tools

System analysis tools

Job description

Lab Summary:

As part of MPS (Mobile Platform and Solutions) and SRA, you will be a member of the XR Core Team focusing on Software Performance Optimization for XR (AR, MR, VR) technologies, building the next generation head mounted devices (HMDs). We work on both commercial and next gen technologies that impact Samsung’s vision for XR. As such you will be working on Software Performance optimizations using the HW/SoC capabilities, meeting the KPIs to launch commercial-level products.

Come join the Samsung MPS team and be a leader in bringing futuristic services to life!

Position Summary:

The XR ML Optimization Engineer will work with the partners, the XR teams in MPS and other Samsung subsidiaries, to successfully launch the XR devices. You will analyze the algorithms and workloads, identifying performance bottlenecks at application, driver and the hardware levels, with the goal of improving the performance, power and SoC utilization. You will identify the key use-cases and decompose them at the system, driver and NPU levels. You will explore and implement more efficient algorithms, driver enhancements, OS optimizations, application-level improvements.

Position Responsibilities:

• Analyze XR workloads to identify bottlenecks in both hardware and software components
• Propose new HW/SW co-optimization methodologies to optimize performance and power of XR devices
• Suggest innovative co-optimization ideas to our software engineers and hardware vendors
• Assist hardware vendors in developing optimal kernels and compilers to process critical XR applications
• Analyze performance estimates for different hardware configurations and kernels
• Work with internal SW engineers to understand their vision and needs from high performance computing cores
• Manage communication and coordination with internal and external partners
• Build simulations and performance models for efficient decision making

Required Skills:

• Bachelor’s or Master’s or PhD in Computer Science/Engineering with specialization in Computer Architecture, Compliers, Parallel Computing, or equivalent combination of education, training, and experience
• 8+ years of experience in software/hardware co-design & optimization with the knowledge of the SoC hardware
• Deep understanding of CPU, GPU, DSP, Deep Learning Accelerators (NSP/NPU) architectures, system programming and optimization of Multimedia/CV/ML algorithms on hardware acceleration cores using C/C++ and Python
• Strong knowledge of computer architecture, memory subsystem, parallel computing and compilers
• Good understanding of ML compute and chip microarchitecture
• Familiarity and hands-on experience with various system analysis tools used for memory, performance analysis and hardware resource management for optimization and stability
• High proficiency in architecture analysis and performance modeling, ranging from simple analytical models to complex cycle accurate performance model and correlation, especially around GPU and NPU
• Expertise in methods for partitioning a solution across hardware and software within an overall systems solution
• Understanding of deep learning algorithms and experience with ML tuning and refinement with ML libraries & frameworks such as PyTorch, Tensorflow, ONNX, Caffe
• Good knowledge of Android/mobile frameworks and Linux/RTOS kernels
• Solid foundation in software development and related tools
• Strong teamwork, communication skills, passion, productivity, and self-learning ability
• Proven ability to work in a dynamic, multi-tasked environment