Edge AI Software Engineer

Description:

Job Summary

The Edge AI Software Engineer is responsible for driving AI capabilities on resource-constrained embedded and edge platforms, enabling the next-generation intelligent evolution of wireless communication modules and IoT products. Beyond traditional embedded and cellular module software, this role focuses on deep integration of AI models with embedded platforms, including MCU, RTOS, and Embedded Linux environments.

This position contributes to the adaptation, optimization, and deployment of AI and large-scale models on constrained devices, balancing compute capability, memory footprint, power consumption, real-time behavior, and system stability.

In addition, the role drives the adoption and production deployment of LLM-based Agent frameworks into daily engineering workflows, improving development efficiency and engineering productivity. The role requires close collaboration with AI algorithm, module platform, system architecture, and product teams to deliver production-ready edge AI solutions.

Objectives \& Responsibilities

• Adapt, optimize, and deploy AI models on embedded and edge platforms, including model pruning, compression, quantization, distillation, and runtime integration for production use
• Design and evolve AI model architectures based on business scenarios and deployment constraints, balancing accuracy, latency, memory usage, power consumption, and system stability on resource-limited platforms.
• Develop and maintain embedded AI software frameworks and inference pipelines, supporting lightweight runtimes such as TensorFlow Lite, TensorFlow Lite Micro, PyTorch Mobile, ONNX Runtime, and similar engines across MCU, RTOS, and Embedded Linux platforms.
• Lead inference acceleration and performance optimization on embedded systems, leveraging platform capabilities and heterogeneous resources (CPU, DSP, NPU, GPU) to continuously improve edge-side AI efficiency.
• Drive the adoption and production integration of LLM-based Agent frameworks, embedding them into daily development and engineering workflows to improve productivity, automation, and system-level efficiency across the SW department and overall R\&D organization.

Requirements:

• Bachelor’s degree or above in Computer Science, Artificial Intelligence, Electronic Engineering, Communications, Automation, or related fields.
• Solid experience in machine learning / deep learning projects, with understanding of large models (LLMs) and end-to-end workflows including training, inference, evaluation, and deployment.
• Proven model engineering experience, capable of advancing models from algorithm validation to deployable, maintainable, and integrable embedded solutions.
• Hands on experience in model structure design and optimization, including pruning, quantization, compression, and distillation based on deployment constraints.
• Practical experience with data preprocessing and feature extraction pipelines, including signal processing techniques such as FFT / STFT, for embedded or edge inference.
• Strong background in embedded AI software frameworks, with experience integrating and adapting at least one of TFLite, TFLite Micro, PyTorch Mobile, ONNX Runtime, or equivalent runtimes.
• Good understanding of MCU, RTOS, and embedded platform architectures, including memory constraints, performance tuning, and power management considerations.
• Proficiency in C / C\+\+, with strong engineering discipline and hands on debugging and optimization experience in embedded environments.
• Good English reading and writing skills, able to read technical documentation and collaborate with global teams.

Preferred Qualifications

• Experience deploying or optimizing LLMs or multimodal models on edge or embedded devices. Experience with Qualcomm based platform will be a strong plus.
• Hands on background in TinyML, edge intelligence, or on device inference projects.
• Experience in one or more of the following areas:
• Lightweight computer vision models on edge devices
• Speech / keyword spotting / time series models on MCU platforms
• Predictive or anomaly detection on embedded systems
• Experience with multi platform inference optimization, including DSP / NPU / GPU acceleration, vectorization, runtime optimization, compiler level or operator level optimization.
• Prior experience with cellular modules, IoT devices, wireless terminals, or embedded products. Understanding of cellular module architecture and modem-host interaction.
• Experience in custom or innovative model architecture design, beyond simply deploying off the shelf models.
• Contribution experience in open-source community.
• Good understanding of the value stream in embedded AI ecosystem.