任職要求 / Qualifications 經(jīng)驗(yàn):5–8 年軟件/數(shù)據(jù)/ML 經(jīng)驗(yàn),2+ 年 LLM/Agentic 實(shí)戰(zhàn),負(fù)責(zé)過(guò)?產(chǎn)級(jí)智能體或 GenAI 應(yīng)?。 5–8 years in SWE/Data/ML, 2+ years hands-on with LLM/agentic systems in production. 技術(shù)棧:精通 Python 或 TypeScript;熟悉 LangGraph/Semantic Kernel/?研編排器 等之?;掌握 FAISS/Milvus/Pinecone/Elasticsearch 任?。 Strong Python or TypeScript; familiar with LangGraph/Semantic Kernel/custom orchestrators; one of FAISS/Milvus/Pinecone/Elasticsearch. ?程能?:具備函數(shù)/?具調(diào)?、結(jié)構(gòu)化輸出(JSON/Schema 校驗(yàn))、多步規(guī)劃/反思/?修復(fù)實(shí)踐;能把 業(yè)務(wù)規(guī)則固化為 Guardrails。 Proven tool/function calling, structured outputs (JSON/schema validation), multi-step planning/reflection/self-repair; encode rules as guardrails. 溝通:英語(yǔ)流利,能與海外客戶(hù)/團(tuán)隊(duì)?效溝通。 Fluent English for effective collaboration with global customers/teams. 加分項(xiàng)(Plus):指令微調(diào)/LoRA/DPO、ServiceNow/Salesforce 集成、數(shù)據(jù)駐留實(shí)踐。 Nice-to-haves: instruction tuning/LoRA/DPO, ServiceNow/Salesforce integrations, data residency practice.