Software Engineer, LiteRT / On-Device AI

Job Title

Software Engineer, LiteRT / On-Device AI

Role Summary

Develop runtime, backend libraries, and optimizations for running AI and LLM models on Qualcomm NPUs. Work across model graph, runtime, memory, and scheduling layers to enable efficient on-device inference.

Collaborate with internal teams, external partners (including open-source communities), and customers to bring models to production on Qualcomm platforms.

Experience Level

Mid-level. The posting is open to a range of experience from new college graduates to experienced engineers; example minimum experience paths listed by the employer include Bachelor's +2 years or Master's +1 year.

Responsibilities

Primary responsibilities include building and optimizing the inference stack for on-device AI and LLMs.

• Enable, bring up, and optimize AI and LLM models on Qualcomm NPU from graph-level transformations to individual ops.
• Design, implement, profile, and analyze runtime, backend, and operation libraries for model execution.
• Diagnose accuracy and performance issues across model, runtime, backend, and hardware to identify root causes.
• Collaborate with partners and open-source communities; deliver high-quality, maintainable code.
• Build and maintain tooling and tests to ensure stability and numeric correctness across platforms.

Requirements

Must-have technical skills and experience. Educational degree requirements are summarized separately below.

• Proficiency in C, C++ and Python for performance-critical software development.
• Strong fundamentals in object-oriented design, data structures, algorithms, operating systems, and computer architecture.
• Understanding of machine learning fundamentals and transformer / LLM architectures (attention, KV cache, tokenization, prefill vs. decode).
• Experience with low-level programming, performance profiling, and optimizing parallel or heterogeneous compute.
• Familiarity with embedded systems constraints and on-device inference considerations (memory, threading, buffer/cache management).
• Proficiency with version control and development tools (e.g., Git, Gerrit) and experience writing tests and tooling for runtime stability.
• Ability to work independently, communicate technical concepts clearly, and collaborate across teams and partners.
• Nice-to-have: familiarity with quantization, ML frameworks/runtimes (TensorFlow, PyTorch, LiteRT), ML compilers (TVM/XLA/Glow), and on-device debugging/profiling tools.

Education Requirements

Employer lists degrees and fields explicitly: Bachelor's, Master's, or PhD in Computer Science, Electrical Engineering, Engineering, Information Systems, or a related field. Example minimum experience paths in the posting: Bachelor's +2 years, Master's +1 year, or PhD (as alternatives tied to degree level).

About the Company

Company: Qualcomm

Headquarters: San Diego, California, United States

Qualcomm is a global leader in semiconductor and telecommunications equipment, specializing in mobile technologies and innovations. Known for its Adreno GPUs, the company provides solutions enabling advancements in mobile gaming, AI, VR/AR, and autonomous driving. Qualcomm's cutting-edge technology and commitment to high-performance, power-efficient designs drive the evolution of mobile graphics and connectivity worldwide.