3 days ago
Bengaluru, IndiaSenior
Responsibilities
- Identify and prioritize ML models and use cases for CPU execution and characterize their compute, memory, cache, parallelism, and dataflow behavior.
- Generate execution traces with QEMU or equivalent simulators and develop tooling to capture instruction behavior, performance counters, and bottlenecks.
- Analyze CPU pipeline, memory hierarchy, and instruction-utilization bottlenecks and optimize hotspots through kernel tuning, algorithmic improvements, and memory optimization.
- Collaborate with CPU architecture and design teams to provide workload insights and propose enhancements to compute units, SIMD extensions, and memory subsystems.
- Design and implement optimized ML kernels and libraries for GEMM, convolution, attention, activation functions, and QMX architectures.
- Integrate kernels with ML frameworks and inference stacks, optimize CPU-centric benchmarks, establish performance baselines, and track improvements across hardware generations.
Requirements
- Bachelor's degree in Engineering, Information Systems, Computer Science, or a related field plus 8+ years of software engineering or related experience; or a master's degree plus 7+ years; or a PhD plus 6+ years.
- At least 4 years of experience with programming languages such as C, C++, Java, or Python.
- Strong computer architecture, systems programming, and machine learning fundamentals.
- Mandatory proficiency in C/C++.
- Experience with performance profiling, benchmarking, and optimization.
- Preferred experience with QEMU or equivalent simulators and ML kernel development for GEMM, convolution, and attention.
- Preferred knowledge of CPU pipelines, caching, SIMD/vector extensions such as NEON, SVE, and QMX.
- Preferred familiarity with ML frameworks, inference stacks, intrinsics, assembly, and memory and cache tuning.
About Qualcomm
Qualcomm is a public semiconductor company headquartered in San Diego, founded in 1985, that designs and sells wireless chipsets and platforms for mobile devices, automotive, IoT, and networking, notably the Snapdragon application processors and 5G modems. It also licenses a large portfolio of cellular patents to device makers, generating revenue alongside chip sales; its technology underpins many Android smartphones and emerging automotive and edge-compute systems, and it trades on NASDAQ as QCOM.