2 days ago
Base Salary
$194k - $292k/yr
Responsibilities
- Lead the bring-up and enablement of LLM, LVM, VLM, and multimodal model architectures on Snapdragon platforms.
- Analyze model architectures, tensor flows, computational graphs, operators, precision, memory behavior, and runtime gaps against Qualcomm AI accelerator backends.
- Design scalable model-enablement architectures, APIs, pipeline stages, configuration schemas, artifact interfaces, and extensibility mechanisms.
- Develop graph adaptation, lowering, decomposition, fusion, layout transformation, quantization, compilation, and on-device execution solutions.
- Build reusable model-enablement frameworks and declarative pipelines for model mapping, quantization recipes, artifact handoffs, caching, validation, and execution workflows.
- Apply post-training quantization, quantization-aware training, mixed precision, weight and activation quantization, calibration, and accuracy-recovery techniques.
- Evaluate model accuracy, latency, throughput, memory usage, model size, and power efficiency across preparation, quantization, compilation, and device execution stages.
- Enable advanced generative AI capabilities including KV-cache optimization, prefill and decode graphs, LoRA, Mixture-of-Experts, speculative decoding, and multimodal processing.
- Debug issues across model architectures, PyTorch and ONNX graphs, quantization, compilers, runtimes, operating systems, and hardware.
- Lead architecture, design, code, performance, and accuracy reviews while establishing technical plans and engineering priorities.
- Manage and mentor a small team of engineers, divide work, manage execution risks, and drive multiple initiatives through production delivery.
- Communicate technical strategy, project status, dependencies, and risks to engineering leadership and cross-functional stakeholders.
Requirements
- A bachelor's degree in computer science, computer engineering, electrical engineering, information systems, or a related field with 8+ years of software design, development, and delivery experience, or a master's degree or PhD with 6+ years of such experience.
- At least 5 years of hands-on experience in AI/ML software development, model inference, model optimization, compiler technologies, or related systems software.
- Strong knowledge of transformer-based LLMs, vision models, multimodal models, and vision-language models, including attention, positional encoding, KV cache, prefill and decode execution, normalization, and feed-forward networks.
- Strong understanding of quantization and calibration, including PTQ, QAT, mixed precision, weight and activation quantization, encoding generation, and accuracy-performance tradeoffs.
- Strong knowledge of computational graphs, graph transformations, operator decomposition, graph lowering, layout and shape transformations, and backend-aware optimization.
- Excellent software architecture and system-design skills with experience designing scalable, modular, extensible, maintainable frameworks, APIs, and production systems.
- Proficiency in Python and C/C++ for production-quality software development.
- Strong experience with PyTorch and ONNX model representations and tooling.
- Experience using agentic coding tools such as Claude Code, OpenAI Codex, or equivalent tools for codebase exploration, implementation, testing, debugging, and code review.
- Demonstrated ability to investigate unfamiliar model architectures, identify backend gaps, and develop scalable technical solutions.
- Strong debugging and root-cause-analysis skills across model, framework, compiler, runtime, and hardware boundaries.
- Demonstrated technical leadership through leading a small team, mentoring engineers, prioritizing work, and driving complex projects end to end.
- Preferred qualifications include experience with embedded, edge, mobile, or accelerator-based platforms; model-enablement pipelines; AIMET; Hugging Face Transformers; PEFT; ONNX Runtime; ExecuTorch; TFLite/LiteRT; MLIR; Android; QNX; Linux; CMake; Git; Snapdragon SoCs; and NPUs or HTPs.
- Preferred experience includes advanced quantization, LoRA, Mixture-of-Experts, speculative decoding, hybrid attention, state-space or recurrent components, vision encoders and decoders, accelerator architecture, and globally distributed team leadership.
Benefits
- The posted annual base pay range is $194,400.00 to $291,600.00, with a competitive annual discretionary bonus program and potential annual RSU grants.
- Qualcomm offers a competitive benefits package designed to support employees at work, at home, and at play.
- The role is posted for a specific location, but the job description does not state a remote, hybrid, or in-office work arrangement.
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.