7 hours ago
Taipei, TaiwanStaff+
Responsibilities
- Optimize training and inference performance across GPU and AI-accelerator infrastructure, including MLOps pipelines.
- Design, train, evaluate, and productionize ML models spanning deep learning, LLMs, computer vision, and recommendation systems.
- Harden and extend NPU cores such as CoralNPU into production silicon.
- Build or optimize inference engines and serving runtimes for real hardware constraints including latency, memory, and power.
- Develop low-level software involving BMC firmware, embedded Linux, or RTOS environments such as Zephyr.
- Build automated test and verification harnesses for AI-assisted RTL/DV, hardware bring-up, and manufacturing test.
- Apply ML to log and intrusion analysis, penetration testing, and firmware or hardware security.
- Collaborate with RTL, hardware, firmware, and QA teams to ship AI features from training through deployment and monitoring.
Requirements
- 5–7+ years of hands-on AI/ML experience.
- Master's degree required; PhD preferred.
- Hands-on experience with AI/ML infrastructure and performance, including GPU clusters, distributed training, inference-serving optimization, and MLOps pipelines.
- Experience designing, training, and evaluating ML models and taking them into production through feature engineering, data pipelines, and deployment.
- Experience optimizing inference engines for specific chips or adapting model architectures and quantization to chip constraints.
- Deep hands-on expertise in at least two specialty areas: NPU/AI accelerators, systems-level software, inference engines/runtimes, test/verification harnesses, or cybersecurity.
- Ability to collaborate across RTL/hardware, firmware, and QA functions.
Benefits
- The company offers a diverse and multifaceted environment with emphasis on collaboration, continuous learning, and mutual support.
- The role is based at the company's headquarters in Silicon Valley.
- Axiado is an equal opportunity employer.
Categories
About Axiado
Axiado builds hardware-anchored platform security for servers and infrastructure, centered on its Trusted Control/Compute Unit (TCU) processor with root-of-trust functions and AI-based threat detection. It sells security silicon, firmware, and reference platforms to server and network equipment makers and cloud providers integrating baseboard management and platform resiliency. The company was founded in 2017, is privately held, and is headquartered in San Jose, California.