Systems Architect: Memory Wall Exploration
About The Position
The Systems Architect: Memory Wall Exploration focuses on understanding how AI execution interacts with platform resources, particularly memory hierarchy, data movement, and shared-memory behavior across heterogeneous compute environments (CPU, GPU, AI accelerators). The architect will engage with internal teams and external partners (including memory and silicon vendors) to translate low-level technical characteristics into system-level guidance and platform strategy for HP. Key Responsibilities Platform Performance & Memory Analysis Analyze how AI workloads interact with system resources across CPU, GPU, and AI accelerators Evaluate the impact of memory bandwidth, latency, contention, and data movement on workload performance, responsiveness, and power efficiency Characterize workload behavior in shared-memory systems, including interactions between foreground and background activity Identify sources of performance variability and experience degradation under real-world multitasking scenarios. Overcome the performance bottleneck where processor speeds outpace memory bandwidth, focusing on optimizing data movement between CPUs, GPUs, and memory (DRAM, HBM, CXL). System-Level Guidance & Strategy Translate memory and system-level constraints into actionable guidance for platform architecture, runtime, power, and performance teams Help define execution envelopes and tradeoffs (e.g., placement, concurrency limits, configuration sensitivity) that improve AI experience predictability Contribute to platform-level strategy by connecting technical constraints to user experience outcomes and product decisions Cross-Functional & Ecosystem Collaboration Work closely with platform architecture, runtime, power/thermal, and AI enablement teams within HP Engage with external ecosystem partners, including memory and silicon vendors, to: Understand platform capabilities, tradeoffs, and roadmap directions Evaluate how memory technologies and configurations affect real-world AI experiences Translate partner insights into HP-specific system guidance and strategic recommendations Support technical alignment discussions Platform Enablement Assist in defining internal tools, metrics, and evaluation methods for assessing AI workload behavior and memory sensitivity Contribute to documentation, best practices, and internal frameworks related to system-level AI performance Help scale AI experiences across a diverse portfolio of devices and SKUs
Requirements
- M.Sc. in Computer Engineering, Electrical Engineering, or a related discipline
- Strong background in systems performance, platform architecture, or runtime behavior
- Solid understanding of memory hierarchy and data movement in client or embedded systems (e.g., DRAM, caches, shared memory architectures)
- Strong knowledge of computer architecture, specifically memory hierarchy (DRAM, SRAM, cache), interconnect protocols (CXL, PCIe), and memory controllers
- Experience with Machine Learning hardware acceleration and processing-in-memory (PIM) techniques
- Knowledge of low-power design, given the "power wall" constraint
- Experience working with heterogeneous compute environments (CPU, GPU, accelerators)
- Ability to translate low-level technical constraints into system-level guidance and platform strategy
- Experience collaborating across hardware, software, and product organizations
- Strong technical communication skills, including engagement with external partners
- Proficiency in C/C++ and scripting languages (Python) for modeling and analysis
- Experience with architectural simulators (e.g., GEM5, Sniper) and performance analysis
Nice To Haves
- Client platforms (PC, mobile, embedded) rather than exclusively datacenter or HPC environments
- On-device AI inference workloads
- Power- and performance-sensitive system design
- Performance analysis tools, telemetry, and data-driven evaluation
- Prior experience engaging with memory or silicon vendors in a technical or architectural capacity
Responsibilities
- Analyze how AI workloads interact with system resources across CPU, GPU, and AI accelerators
- Evaluate the impact of memory bandwidth, latency, contention, and data movement on workload performance, responsiveness, and power efficiency
- Characterize workload behavior in shared-memory systems, including interactions between foreground and background activity
- Identify sources of performance variability and experience degradation under real-world multitasking scenarios
- Overcome the performance bottleneck where processor speeds outpace memory bandwidth, focusing on optimizing data movement between CPUs, GPUs, and memory (DRAM, HBM, CXL)
- Translate memory and system-level constraints into actionable guidance for platform architecture, runtime, power, and performance teams
- Help define execution envelopes and tradeoffs (e.g., placement, concurrency limits, configuration sensitivity) that improve AI experience predictability
- Contribute to platform-level strategy by connecting technical constraints to user experience outcomes and product decisions
- Work closely with platform architecture, runtime, power/thermal, and AI enablement teams within HP
- Engage with external ecosystem partners, including memory and silicon vendors, to understand platform capabilities, tradeoffs, and roadmap directions
- Evaluate how memory technologies and configurations affect real-world AI experiences
- Translate partner insights into HP-specific system guidance and strategic recommendations
- Support technical alignment discussions
- Assist in defining internal tools, metrics, and evaluation methods for assessing AI workload behavior and memory sensitivity
- Contribute to documentation, best practices, and internal frameworks related to system-level AI performance
- Help scale AI experiences across a diverse portfolio of devices and SKUs
Benefits
- Health insurance
- Dental insurance
- Vision insurance
- Long term/short term disability insurance
- Employee assistance program
- Flexible spending account
- Life insurance
- Generous time off policies, including; 4-12 weeks fully paid parental leave based on tenure
- 11 paid holidays
- Additional flexible paid vacation and sick leave (US benefits overview)
Stand Out From the Crowd
Upload your resume and get instant feedback on how well it matches this job.
What This Job Offers
Job Type
Full-time
Career Level
Senior
