Senior AI Software Engineer (Agentic Engineering)

工作說明

1. Multi-Agent Architecture Design & Implementation: Architect a Multi-Agent collaboration system from the ground up, defining inter-agent state management, domain knowledge access mechanisms, and behavioral conventions (Conventions). Establish foundational agent reflection mechanisms (Reflections) and standard operating templates, and drive core proof-of-concept (POC) development through to iterative production deployment.
2. LLM Framework & Routing Integration: Evaluate and adopt the most suitable AI development toolkits for our stack. Implement and maintain routing tools such as LiteLLM to unify our internal multi-model API call architecture—optimizing for token efficiency, response latency, and high availability.
• Context Engineering: Design and optimize RAG pipelines and knowledge extraction workflows tailored to our existing codebase and technical documentation, ensuring AI outputs remain accurate, grounded, and low-hallucination within the highly specialized business context of medical systems.
3. Frontier AI Architecture Validation & Operationalization: Continuously benchmark emerging AI paradigms—including Recursive Agent Architecture, Agent SKILL frameworks, and MCP—and translate the most promising into tangible engineering tools that elevate team-wide productivity.

需求條件

1. 5+ years of software development experience with strong Python engineering fundamentals.
2. Proficiency in at least one of: Claude Code, OpenAI Codex, or GitHub Copilot.
3. Strong Software Architecture Design skills, with the ability to clearly articulate and plan the underlying logic of complex systems.
4. Hands-on production experience with LLM integration frameworks and routing tools; proficiency in LangChain, Google Agent Development Kit (ADK), or LiteLLM is required.
5. Solid expertise in Context Engineering, with practical experience building LLM prompt pipelines or RAG architectures.
6. A hacker's mindset for hard problems—energized by unknown technical bottlenecks and able to systematically break them down.
7. Exceptional clarity of logical thinking, with the ability to explain complex architectures to non-technical stakeholders in plain language.
8. Hands-on experience with Kubernetes (K8s) containerized deployments, microservice architecture, and Linux system administration.
9. Solid foundations in networking and low-level infrastructure knowledge.
10. Proficiency in JavaScript / Vue.js for rapidly building frontend validation prototypes.