Senior Performance Modeling Engineer

Do you want to help shape the future of AI infrastructure and influence the hardware-software co-design decisions that power Microsoft's next generation AI platforms?

Join the Systems Planning and Architecture (SPARC) team within Azure Hardware Systems and Infrastructure (AHSI), where we are building the foundation for the world's largest AI workloads. Our team drives model enablement, performance optimization, and hardware-software co-design across Microsoft's custom AI accelerators and cloud-scale AI infrastructure.

We are seeking a Senior AI Performance Modeling Engineer to join the AI Systems Architecture (ASA) group within Microsoft SPARC, where we define and optimize next-generation AI accelerators and large-scale AI systems. In this role, you will develop analytical performance models, simulation frameworks, and workload characterization tools to evaluate accelerator and system architectures for large-scale AI training and inference.

You will model real-world AI workloads across compute, memory, and communication subsystems, identifying performance bottlenecks, scalability limits, and architectural trade-offs. A key aspect of this role is evaluating new hardware features, parallelization strategies, and system configurations to optimize performance, power efficiency (perf/W), and total cost of ownership (TCO).

You will collaborate closely with architecture, microarchitecture, compiler, runtime, and systems teams to correlate analytical predictions with silicon measurements and software traces, improve modeling accuracy, and drive data-driven architectural decisions for future AI platforms.

This role is flexible and offers a hybrid work model with three days per week in the office.

Microsoft's mission is to empower every person and every organization on the planet to achieve more. As employees, we come together with a growth mindset, innovate to empower others, and collaborate to achieve our shared goals. Each day we build on our values of respect, integrity, and accountability to create a culture of inclusion where everyone can thrive at work and beyond.

• Develop and enhance analytical performance models and simulation frameworks for AI accelerators, GPU architectures, and large-scale AI systems.

• Model end-to-end AI training and inference workloads, including compute kernels, attention, MoE, memory access patterns, and communication operations.

• Evaluate new architectural features, compute and memory hierarchies, interconnects, and collective communication mechanisms to identify performance and efficiency improvements.

• Perform workload mapping and parallelization analysis across tensor, pipeline, expert, data, and context parallelism strategies to optimize accelerator utilization and scalability.

• Develop system-level performance models to evaluate throughput, latency, perf/W, and TCO across accelerator, rack, and cluster configurations.

• Correlate analytical model predictions with silicon measurements, kernel benchmarks, and software traces to validate assumptions, improve model fidelity, and identify performance bottlenecks.

• Build and automate performance modeling, design-space exploration, sensitivity analysis, and visualization tools to accelerate architectural evaluation.

• Collaborate with hardware and software teams to translate modeling insights into actionable architecture recommendations and HW–SW co-design opportunities.

• Doctorate in Electrical Engineering, Computer Engineering, Mechanical Engineering, or related field AND 3+ years technical engineering experience OR Master's Degree in Electrical Engineering, Computer Engineering, Mechanical Engineering, or related field AND 6+ years technical engineering experience OR Bachelor's Degree in Electrical Engineering, Computer Engineering, Mechanical Engineering, or related field AND 8+ years technical engineering experience OR equivalent experience
• Familiarity with AI agentic systems & flow for performance modeling & anlaysis
• 2+ years of experience in Computer Architecture, AI Systems, or closely related technical domains.
• Strong understanding of GPU and AI accelerator architectures, including compute pipelines, memory hierarchies, interconnects, and parallel execution models.
• Experience developing analytical performance models, roofline models, architectural simulators, or system-level performance estimation frameworks.
• Knowledge of AI training and inference workloads, including transformers, attention mechanisms, MoE architectures, and large-scale distributed execution.
• Experience modeling compute, memory bandwidth, communication collectives, and computation-communication overlap in accelerator-based systems.
• Familiarity with parallelization strategies, including tensor, expert, pipeline, data, and context parallelism, and their performance implications.
• Strong programming skills in Python and C/C++, with experience developing modeling frameworks, simulators, and performance analysis tools.
• Experience with performance profiling, benchmarking, and correlation of analytical models with hardware measurements and software traces.
• Familiarity with PyTorch, GPU programming, and modern AI inference or training frameworks such as vLLM, SGLang, and distributed runtime systems.
• Experience evaluating architectural design trade-offs using performance, power efficiency, scalability, and TCO metrics.
• Strong analytical, problem-solving, and communication skills, with the ability to translate performance insights into architectural recommendations.