1 day ago
Hsinchu, Taiwan or Taipei, TaiwanEntry Level / Mid Level
Responsibilities
- Develop core runtime and backend libraries that execute AI and LLM workloads on Qualcomm NPUs.
- Enable, bring up, and optimize AI, LLM, and generative models through graph transformations and operation-level optimization.
- Design, implement, profile, and analyze runtime, backend, and operation libraries for performance-critical inference.
- Diagnose accuracy and performance problems across models, runtimes, backends, and hardware and drive issues to root cause.
- Build and maintain tooling and test coverage for runtime stability and numerical correctness across platforms.
- Collaborate with Qualcomm teams, open-source communities, industry partners, customers, and technical specialists to improve libraries for commercial use cases and benchmarks.
Requirements
- A bachelor's degree, master's degree, or PhD in Computer Science, Engineering, Information Systems, or a related field, with experience requirements varying by degree path.
- Proficiency in C, C++, or Python; experienced candidates should have strong modern C++ and Python skills.
- Knowledge of object-oriented programming, data structures, algorithms, operating systems, computer architecture, and parallel programming.
- Understanding of machine learning, deep learning, neural networks, transformers, LLMs, attention, KV cache, tokenization, and prefill versus decode.
- Understanding of NPU, DSP, and GPU architectures, embedded systems, edge-device constraints, and hardware utilization optimization.
- Experience or familiarity with performance analysis, low-level programming, memory and threading behavior, resource scheduling, quantization, ML frameworks, runtimes, and model formats.
- Familiarity with Git, Gerrit, Jira, and tools for debugging and profiling on-device code.
- Preferred experience includes LiteRT, TensorFlow, PyTorch, LLM or multimodal model enablement, ML compilers such as TVM, XLA, or Glow, and LLVM or GCC backend development.
- Ability to collaborate with partners and customers, communicate technical concepts clearly, work independently, and resolve technical problems.
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.