Senior or Staff Software Engineer — QAIRT Windows Platform

Qualcomm is a world leader in on-device AI, delivering the hardware and software that powers intelligent experiences across mobile, automotive, IoT, and PC platforms.

Taiwan is not a satellite office — it is where the Windows platform is built. Our AI Software Engineering team here is Qualcomm's sole core development team for the QAIRT SDK on Windows, holding end-to-end ownership of the platform — not as a downstream integrator, but as the primary engineering driver. Part of Qualcomm's HTP core engineering team is co-located with us in Taipei, and we work in tight collaboration with the common platform teams in San Diego and Markham who serve mobile, automotive, and IoT. If you want to work on something that matters to the whole company's AI PC strategy, and do it from Taiwan, this is that opportunity.

The scope of the role spans the entire QAIRT SDK surface on Windows on Snapdragon: from developer-facing sample tools and APIs, through the runtime inference pipeline, all the way down to the HTP backend interfacing with Qualcomm's Hexagon neural network engine. The team owns runtime stability, inference performance, feature parity, and long-term software compatibility — ensuring that applications built against the SDK continue to work correctly as Windows versions, SoC generations, and driver stacks evolve independently over time. This last point is not routine maintenance: it is an active design discipline, requiring careful thinking about API contracts, version negotiation, and failure modes that only surface when software that was never tested together ends up running together in the field.

The Role

Senior / Staff Software Engineer on the QAIRT Windows platform team, responsible for the full-stack development and long-term health of the QAIRT SDK on Windows on Snapdragon.

The Person

You are a senior systems software engineer who can own complex problems from first principles to production. You navigate large, multi-layered C++ codebases comfortably, reason about the full execution path from application API down to hardware-accelerated inference, and have the instincts to know when a problem is a bug, a design flaw, or a fundamental platform constraint.

You have strong opinions about software quality — not just whether code works today, but whether it will still work in three years when the device firmware, the OS, and the application have all been updated independently. You communicate well across team boundaries, can represent technical decisions to global stakeholders, and are the kind of engineer others come to when something is broken and nobody knows why.

We are a small team with a wide scope. Engineers here are expected to own problems fully — from the first crash log to the design review to the fix in production — and to push back when something is wrong, regardless of where the issue originated. There is no one to hand the hard problems off to.

Key Responsibilities

• Own the development and long-term health of the QAIRT SDK on Windows on Snapdragon, across the full stack from developer-facing APIs and sample tools to the HTP backend interfacing with Qualcomm's Hexagon neural network engine

• Investigate and resolve runtime failures, inference accuracy issues, and performance regressions surfaced through internal validation, partner integration, or production deployments

• Port and enable new features and operator support across SDK releases, ensuring correctness and performance on Qualcomm hardware acceleration cores

• Design SDK components for durability — ensuring the stack behaves correctly when different versions of the SDK, SoC driver, and Windows OS are combined in the field, including combinations that were never explicitly tested together

• Drive technical decisions and architectural discussions in collaboration with the HTP core team, the common platform teams in San Diego and Markham, and hardware and driver stakeholders

• Contribute to the testing and tooling infrastructure that gives the team confidence to ship — covering compatibility scenarios, performance baselines, and runtime correctness

Minimum Qualifications

• Bachelor’s degree in Computer Science, Electrical Engineering, or related field

• 5+ years of software engineering experience in systems software, SDK development, or a related discipline

• Proficiency in C/C++ with a focus on Windows platform development

• Experience in large-scale software projects with a strong emphasis on correctness, stability, and maintainability

• Proficiency with version control and code review tools such as Git and Gerrit

• Solid understanding of system architecture and software design principles, including API design, versioning, and cross-component compatibility

• Ability to communicate technical concepts clearly and work effectively across distributed, cross-functional teams

Preferred Qualifications

• 8+ years of software engineering experience in the semiconductor, systems software, or consumer devices industry, with a track record of owning significant components end-to-end

• Strong foundation in software design patterns, performance engineering, and building systems that remain correct under long-term evolution

• Proven ability to debug and resolve deep runtime or system-level issues — the kind that require reasoning across multiple layers of the stack simultaneously

• Experience with Windows user-space development, including diagnostic and debugging tools such as WinDbg, Application Verifier, or dumpbin

• Experience developing or integrating AI inference SDKs, runtime libraries, or compiler backends at production quality

• Familiarity with AI frameworks such as PyTorch or ONNX Runtime, particularly in deployment or productization contexts rather than model training

• Understanding of performance profiling and optimization for compute-intensive workloads on heterogeneous hardware (CPU, DSP, GPU)

• Demonstrated experience incorporating AI coding assistants (such as Claude Code, ChatGPT, Codex, or Gemini) into day-to-day development workflows — including code generation, debugging, and code review — with a clear understanding of how to validate and take ownership of AI-generated output

Expertise in at Least One of the Following Areas

• Windows ecosystem development:
  Deep familiarity with Windows user-space internals, DLL loading and versioning, system compatibility mechanisms, and diagnostic tooling such as WinDbg, Application Verifier, or dumpbin

• Performance profiling and optimization:
  Practical experience identifying and eliminating bottlenecks across CPU, memory, and hardware accelerator pipelines, with measurable impact on production workloads

• AI and deep learning fundamentals:
  Working knowledge of DNN architectures and inference optimization techniques, with hands-on experience in PyTorch or ONNX Runtime in deployment or productization contexts

• Low-level SDK or runtime development:
  Experience building system-level libraries or inference runtimes where API stability, binary compatibility, and cross-version correctness are first-class requirements