About the role#
As an intern on the Machine Learning Systems team, you will work at the intersection of machine learning, compiler technology, and AI hardware architecture. You will collaborate with software, firmware, and hardware teams to improve how AI models run on Qualcomm Neural Processing Units (NPUs). This role involves developing production compiler technology and building tools to evaluate future AI architectures. You will gain experience in hardware and software co-design while analyzing performance, power efficiency, and memory usage for workloads like large language models, generative AI, and computer vision.
What you'll need#
- Pursuing a bachelor's degree in Computer Engineering, Electrical Engineering, Computer Science, or a related field.
- Availability for a 16-month internship from May/June 2027 to August/September 2028, with an expected graduation date of November 2028 or later.
- Proficiency in Python or C++.
- Strong analytical, problem-solving, and debugging skills.
- Strong written and verbal communication skills.
- Detail-oriented approach to complex technical problems.
- Familiarity with object-oriented design principles.
- Experience with software development tools including Git, CI/CD systems, Linux, and SSH.
- Understanding of algorithms and data structures.
- Coursework, research, or project experience in machine learning frameworks like PyTorch, ONNX, TensorFlow, or JAX.
- Knowledge of model optimization, AI accelerator architecture, computer architecture, memory systems, and high-performance computing.
What you'll do#
- Compile and optimize neural networks for Qualcomm AI accelerators.
- Analyze model performance to improve compiler algorithms, runtime efficiency, and hardware utilization.
- Prototype algorithms to help AI workloads run with high performance and low power consumption.
- Identify compute and memory hierarchy bottlenecks.
- Recommend software and hardware enhancements to improve system performance.
- Contribute to the Qualcomm ML compiler stack.
- Explore and prototype optimization techniques for emerging neural network architectures.
- Analyze ML workloads to guide future hardware and software development.
- Develop performance models and simulation infrastructure using Python, C, and C++.
- Evaluate current and future Qualcomm AI accelerators to predict workload performance.
Location & details#
- Location: Markham, Ontario, Canada.
- Duration: 16 months.
- Modality: On-site.
- This is a paid internship.


