3 days ago
Beijing, ChinaSenior
Responsibilities
- Deploy, integrate, and validate AI/ML and LLM inference workloads on Qualcomm Cloud AI accelerator platforms.
- Support customer model onboarding, environment setup, functional validation, benchmarking, issue reproduction, and production-readiness activities.
- Debug issues across applications, frameworks, runtimes, drivers, operating systems, and platform interfaces while collecting logs and traces.
- Create and maintain Python and Bash automation for deployment, regression testing, benchmarking, log collection, and result analysis.
- Analyze basic LLM performance using latency, throughput, tokens per second, TTFT, and resource-utilization metrics.
- Support single-server and basic multi-accelerator or multi-node validation, including Scale-Up, Scale-Out, AllReduce, and AllGather concepts.
- Collaborate with model, runtime, compiler, driver, firmware, platform, and infrastructure teams to identify ownership and track resolution.
- Document setup procedures, known issues, troubleshooting guides, benchmark results, and customer enablement practices.
Requirements
- A bachelor's degree in Engineering, Information Systems, Computer Science, or a related field, or a master's degree, PhD, or equivalent practical experience.
- At least 2 years of relevant software applications or software development experience with a bachelor's degree, at least 1 year with a master's degree, or no stated experience requirement with a PhD.
- At least 1 year of academic and/or work experience with programming languages such as C, C++, Java, or Python.
- At least 1 year of academic and/or work experience with debugging techniques.
- Solid software development and debugging skills in Python and/or C/C++, plus practical Bash and Linux command-line experience.
- Understanding of Linux software environments, processes, memory, networking basics, logs, packages, containers, and system troubleshooting.
- Practical exposure to AI/ML inference and at least one framework such as PyTorch, ONNX Runtime, or Hugging Face Transformers.
- Basic understanding of Transformer and LLM concepts, including tokens, attention, context length, KV cache, prefill/decode, and inference performance metrics.
- Experience with software integration, validation, test automation, issue reproduction, and root-cause analysis across multiple software components.
- Ability to analyze technical problems methodically and communicate findings clearly with engineering teams and customers.
- Comfort working directly with customers or partners in evaluation, integration, or technical support environments.
- Preferred qualifications include experience with Hugging Face, vLLM, SGLang, DeepSpeed, model optimization, Docker, Kubernetes, multi-accelerator systems, RDMA/RoCE/InfiniBand, profiling, telemetry, regression analysis, or Linux-based compute platforms.
Categories
Solutions Engineering
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.