Onsite
Who We Are
This role is part of the Kinara Group at NXP Semiconductors, combining the agility and innovation of a startup with the scale, stability, and global reach of a leading semiconductor company.
Kinara was founded on patented technology developed during the founders' Ph.D. research at Stanford University, with a mission to redefine AI inference at the edge through tightly integrated hardware and software. Following Kinara's acquisition, this mission continues within NXP, with the Kinara team operating as a dedicated AI accelerator and software group focused on pushing the boundaries of edge AI.
The team has delivered two generations of AI silicon shipping in volume, supported by a mature and production-ready software stack, making it one of the most proven edge AI platforms in the industry. Kinara technology is used by marquee customers, including GAFA-class companies, a leading global e-commerce provider, and Tier-1 PC OEMs.
As part of NXP, the Kinara Group now benefits from world-class silicon execution, manufacturing scale, and customer access, while retaining its startup culture, deep AI expertise, and compiler-driven engineering mindset. Joining this team offers the best of both worlds: the opportunity to work on cutting-edge AI compiler and silicon technology, with real-world impact and long-term stability.
What We Do
We build game-changing edge AI solutions that enable faster, smarter, and more power-efficient decision making where every millisecond matters.
At the core of our platform are Ara AI inference processors, purpose-built to deliver unrivaled deep learning performance per watt at the edge. Paired with our comprehensive SDK and software toolchain, these processors enable developers to deploy and optimize complex AI models for real-time inference under tight latency and power constraints.
Kinara's hardware-software co-design approach allows high-performance AI to be seamlessly embedded into edge devices, enabling applications across vision, industrial automation, consumer electronics, and intelligent systems. Our solutions focus on performance, efficiency, and scalability, helping customers bring advanced AI capabilities out of the cloud and into the real world.
Edge AI is at an inflection point, and the Kinara Group within NXP is uniquely positioned to play a leading role in its growth, delivering production-proven silicon, a robust compiler and SDK stack, and a clear roadmap for next-generation intelligent edge devices.
This is what you are responsible for
Implement compiler modules within defined subsystems: MLIR dialects, lowering passes, operator fusion strategies, and target code generation components for our NPU.
Participate in design and deployment of quantization flows and numerical-correctness verification for assigned features.
Develop and ship production-grade compiler passes and runtime integration components that are stable, debuggable, and performant.
Perform low-level optimizations and collaborate with kernel teams to address HW/SW performance gaps.
Contribute to compiler CI pipelines, test suites, profiling, and release processes.
Debug and root-cause performance and correctness issues on customer models; produce reproducible fixes.
Mentor junior engineers and contribute to design reviews.
Provide feedback to product and silicon teams on compiler-driven feature opportunities and tradeoffs.
Necessary Qualifications
7-8 years of software engineering experience with hands-on compiler and systems work.
Experience with graph lowering, IR design, operator fusion, scheduling, and code generation for accelerators.
Strong systems programming skills -- proficient C++ and solid Python for tooling and infrastructure.
Track record of shipping compiler components or systems software used in deployed products.
Solid debugging and profiling skills with common performance toolchains.
Strong communication skills and ability to work cross-functionally with hardware architects and kernel teams.
Willingness to mentor junior engineers and contribute to team engineering practices.
Good to Have
Familiarity with MLIR-based frontends and dialects, such as TOSA, StableHLO, TorchMLIR.
Exposure to AI compiler stacks, hardware backends, or accelerator targeting.
Experience working with large-scale models or production ML inference pipelines.
Want to know if this job is worth applying to?