锦鲤求职岗位描述
ORGANIZATION: Global Digital Technologies
ABOUT THE ROLE
We are looking for AI Engineers who can design, build, and deploy autonomous agentic systems that operate reliably in production environments.
This is not a prompt-engineering role
This is not a research-only role
You will architect, build and deploy end-to-end:
• Agentic systems that perform complex workflows, e.g. plan and execute multi-step tasks
• Tool-using LLM systems (retrieval, function calling, code execution, browsing, etc.)
• Multi-agent orchestration frameworks
• Long-term memory architectures (vector + structured state)
• Evaluation harnesses for reliability, hallucination detection, and safety
• Production inference pipelines (latency, scaling, observability)
• Guardrails and fallback systems
RESPONSIBILITIES
• Design and build production-grade agentic AI systems end-to-end, including planning and execution loops, stateful memory, secure tool invocation, sandboxed execution, and robust retry and reflection mechanisms
• Deploy and operate scalable inference infrastructure optimized for latency and cost, with strong
observability through logging, tracing, evaluation metrics, and proactive monitoring for drift and failure modes
• Establish rigorous reliability and safety frameworks by developing automated evaluation pipelines,
product-aligned benchmarks, and stress tests for reasoning performance.
• Implement guardrails and governance controls, including constraint systems, hallucination mitigation, permissioning, audit trails, and human-in-the-loop workflows to ensure secure and dependable real-world operation.
QUALIFICATIONS
• PhD in Computer Science, Mathematics, Engineering, or a related field preferred; outstanding candidates with a Master's degree will also be considered. All candidates must have hands-on experience in building multi-agent systems and/or conducting research in LLM, agents, or related areas.
• Strong hands-in proficiency with AI-native development tools such as Claude Code, Cursor or comparable AI coding assistants, with demonstrated ability to use them to accelerate SDLC
• Deep understanding of tool calling / function calling, RAG architectures, Vector databases, Prompt chaining vs planner/executor models, Latency optimization and cost tradeoffs
• Experience deploying cloud-native systems
• Familiarity with observability stacks (metrics, tracing, logging) and production monitoring practices
• Contributors to open-source AI tooling are strongly preferred
• Example tech stack: Python, FastAPI, LLM APIs and open-weight models, Vector databases, Kubernetes /serverless deployment