Qualcomm develops technologies in AI, high-performance computing, and connectivity to enable intelligent computing and engineering human progress. The Machine Learning Compiler & Performance Engineering Intern will work with software, firmware, performance architecture, and hardware teams to optimize AI model execution on Qualcomm Neural Processing Units, contribute to ML compiler development, and build performance models and simulation infrastructure.
Responsibilities
- Compile and optimize neural networks for execution on Qualcomm AI accelerators
- Analyze model performance and identify opportunities to improve compiler algorithms, runtime efficiency, and hardware utilization
- Prototype algorithms that enable AI workloads to execute with high performance and low power consumption on Qualcomm hardware
- Identify bottlenecks in compute and the memory hierarchy
- Recommend software and hardware enhancements to improve overall system performance and efficiency
- Contribute to Qualcomm's ML compiler stack to improve performance, efficiency, scalability, and developer productivity
- Explore and prototype optimization techniques for emerging AI workloads and state-of-the-art neural network architectures
- Analyze ML workloads to help guide future hardware and software feature development
- Develop performance models and simulation infrastructure using Python, C, and C++
- Evaluate current and future Qualcomm AI accelerators and predict workload performance before silicon is available