3 months ago
Taipei, TaiwanStaff+
Responsibilities
- Act as a senior individual contributor and technical architect through hands-on coding, design, analysis, and technical leadership across the ML stack.
- Define end-to-end architecture for MLOps, agentic AI, model optimization, and security strategies.
- Design data processing, data versioning, labeling, active-learning, and production monitoring pipelines.
- Build human-in-the-loop and AI-in-the-loop systems to continuously improve datasets.
- Develop lightweight on-device monitoring for operational metrics, inference quality, and concept drift.
- Design and develop autonomous agents for resource-constrained edge devices using log analysis, computer vision, and open-source system tools.
- Implement model security and verification capabilities including secure protocols, authentication, model signing, and model injection prevention.
- Bridge ML software development with silicon teams and collaborate with RTL designers on NPU and FPGA architecture.
- Lead model optimization research using graph-level techniques such as operator fusion and operator-level techniques such as custom operations.
Requirements
- 8–10+ years of hands-on machine learning experience with a senior- or staff-level individual contributor track record.
- Ph.D. or M.S. in Computer Science, Electrical Engineering, or a related field, or equivalent practical experience.
- Expert-level Python programming and deep experience with ML frameworks such as PyTorch and TensorFlow.
- Deep theoretical understanding of modern machine learning algorithms such as Transformers.
- Foundational understanding of computer architecture, digital logic, and RTL, including Verilog/VHDL's role in hardware design.
- Experience architecting and building complete MLOps lifecycles from data ingestion through production monitoring and labeling loops.
- Experience developing agentic systems or applications using LLMs.
- Domain knowledge in log analysis and/or computer vision.
- Experience with on-device model security, verification, anti-injection techniques, and secure communication protocols.
- Hands-on experience optimizing models for NPUs or GPUs at graph and operator levels.
- Preferred experience with ML compilers such as Apache TVM or MLIR.
- Preferred hands-on experience with Kubernetes for MLOps, including Kubeflow or Argo.
- Preferred familiarity with GPU scheduling and virtualization platforms such as Run:AI.
- Preferred embedded systems experience with C++, Rust, or Yocto.
- Preferred familiarity with the RISC-V instruction set architecture.
- Preferred proficiency with AWS, GCP, or Azure.
Benefits
- The company describes a diverse, collaborative, continuous-learning environment with mutual support.
- The role is based at the company headquarters in Silicon Valley; no remote or hybrid arrangement is stated.
- Axiado is an Equal Opportunity Employer.
Tech Stack
Categories
