Cloud and AI System Intern

Job Details: Job Description: In this position, you will work with a system reliability research team focusing on RAS (Reliability, Availability, Serviceability) and silent data error (SDE) characterization and mitigation on AI and general-purpose compute platforms, including heterogeneous systems (CPU + GPU/accelerators) and large-scale server clusters. You will help design and run experiments under representative AI training/inference and cloud workloads, analyze fleet-scale logs/telemetry, and prototype detection/diagnosis methods to improve end-to-end data integrity and platform robustness across the HW/FW/OS/runtime stack. Your responsibilities will include but not be limited to: -Collect, clean, and analyze platform telemetry / error logs from CPU servers and accelerator-enabled nodes (e.g., memory/DDR/HBM, storage, interconnect, PCIe/CXL, fabrics) to identify error signatures and failure patterns. -Design and execute fault injection, stress tests, or workload-driven experiments to reproduce silent data corruption scenarios for AI training/inference and general compute workloads, and validate hypotheses. -Research and analyze in-field scan and lockstep mode features (coverage, limitations, trigger conditions, and impact on AI/CPU workloads), and help evaluate how they can be leveraged to improve silent error detection and data integrity in production. -Research and analyze Silicon Lifecycle Management (SLM) solutions, and integrate them with platform telemetry to enable in-field health monitoring, degradation/trend analysis, and proactive reliability improvements for AI/CPU platforms. -Develop scripts/tools (Python preferred) to automate data processing, experiment orchestration, and report generation; build dashboards or repeatable pipelines when needed. -Study and evaluate mitigation techniques for AI + CPU platforms (e.g., ECC/CRC/EDAC, scrubbing policies, retry/recovery, checkpoint/restart, end-to-end checks at data/communication boundaries) and quantify effectiveness vs. performance/cost impact. -Collaborate with cross-functional teams (HW, FW, OS, driver/runtime, datacenter operations) to trace error propagation paths and drive actionable improvements; document findings and present progress regularly. Qualifications: Cloud and AI System Engineering Intern Description In this position, you will work on a system reliability research topic with platform engineering team focusing on RAS (Reliability, Availability, Serviceability) and silent data error (SDE) characterization and mitigation on AI and general-purpose compute platforms, including heterogeneous systems (CPU + GPU/accelerators) and large-scale server clusters. You will help design and run experiments under representative AI training/inference and cloud workloads, analyze fleet-scale logs/telemetry, and prototype detection/diagnosis methods to improve end-to-end data integrity and platform robustness across the HW/FW/OS/runtime stack. Your responsibilities will include but not be limited to: -Collect, clean, and analyze platform telemetry / error logs from CPU servers and accelerator-enabled nodes (e.g., memory/DDR/HBM, storage, interconnect, PCIe/CXL, fabrics) to identify error signatures and failure patterns. -Design and execute fault injection, stress tests, or workload-driven experiments to reproduce silent data corruption scenarios for AI training/inference and general compute workloads, and validate hypotheses. -Research and analyze in-field scan and lockstep mode features (coverage, limitations, trigger conditions, and impact on AI/CPU workloads), and help evaluate how they can be leveraged to improve silent error detection and data integrity in production. -Research and analyze Silicon Lifecycle Management (SLM) solutions, and integrate them with platform telemetry to enable in-field health monitoring, degradation/trend analysis, and proactive reliability improvements for AI/CPU platforms. -Develop scripts/tools (Python preferred) to automate dat ... (truncated, view full listing at source)