
AI Engineer (內湖瑞光)
Advantech Co., Ltd.4 days ago
Taipei, TaiwanMid Level
Responsibilities
- Fine-tune open-weight LLMs using supervised fine-tuning, LoRA, QLoRA, DPO, and RLHF where appropriate.
- Curate, clean, label, synthesize, version, and maintain training and evaluation datasets with domain experts.
- Build benchmarks and evaluation sets comparing fine-tuned models with base models and prompt-only or RAG baselines.
- Package, deploy, and optimize models through the NemoClaw model gateway on on-premises GPUs and in the cloud.
- Design tool and function schemas with structured outputs and implement reliable tool-calling logic with validation, retries, and error handling.
- Build MCP servers, reusable skills, multi-step agent workflows, and A2A hand-offs.
- Integrate agents with IoTMart, Data Lake, SFDC/CRM, and other enterprise systems through governed APIs.
- Apply least-privilege access, output validation, prompt-injection defenses, and human approval for high-impact or write actions.
- Trace model and tool calls, run evaluation-driven regression tests, and monitor production accuracy, latency, and token cost.
- Build retrieval pipelines and combine RAG, tools, and fine-tuned models for enterprise use cases.
- Develop backend services, APIs, chat interfaces, admin consoles, and other full-stack applications.
- Implement the AI Architect’s platform architecture and share fine-tuning recipes, evaluation harnesses, and tool or MCP templates with the wider team.
Requirements
- At least 3 years of software-engineering experience, including at least 2 years of hands-on LLM-application development.
- Hands-on experience preparing datasets, fine-tuning open-weight LLMs with supervised and parameter-efficient methods, and evaluating results against baselines.
- Hands-on experience designing tool or function schemas and building multi-step agents that call real APIs with structured outputs, validation, and error handling.
- Experience creating evaluation sets and metrics for LLM or agent behavior and using results to guide engineering decisions.
- Strong Python and software-engineering fundamentals, including version control, testing, and CI/CD.
- Full-stack capability across backend services, APIs, and front-end development.
- Experience deploying and serving models or AI services using containers and GPU inference in Azure, AWS, and/or on-premises environments.
- Working knowledge of retrieval-augmented generation and vector search.
- Bachelor’s or Master’s degree in Computer Science, Engineering, or a related field, or equivalent practical experience.
- Professional working proficiency in English and Mandarin Chinese.
- Preferred experience with DPO, RLHF, distillation, synthetic data generation, MCP servers, agent frameworks, and A2A patterns.
- Preferred experience with quantization, vLLM, TensorRT-LLM, NVIDIA AI infrastructure, LLM observability, evaluation tooling, token-cost management, and security for tool-using agents.
- Preferred end-to-end full-stack delivery experience using React or Vue with Python or Node.js backends.
- Preferred knowledge of CRM/SFDC, ERP, e-commerce systems, and their APIs, plus public models, open-source contributions, or project write-ups.
Categories
About Advantech Co., Ltd.
Advantech Co., Ltd. designs and sells industrial PCs, embedded boards, IoT gateways, and edge-computing platforms—plus software such as its WISE-IoT suite—for manufacturers, utilities, healthcare, transportation, and retail. The company also provides design-in and systems integration services. Founded in 1983 and headquartered in Taipei’s Neihu District, it is a public company listed in Taiwan and serves customers globally.