Hire this AI Trainer
Sign in or create an account to invite AI Trainers to your job.
Text
Prompt + Response Writing (SFT)
Function Calling上海塞舵智能科技有限公司 LLM运营实习生（构建评测与错误标签体系）. Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal, Proprietary Tooling, and Other. Education includes Bachelor of Science, Donghua University (2024) and Master of Science, Donghua University (2027). AI-training focus includes data types such as Text and labeling workflows including Evaluation, Rating, and Prompt + Response Writing (SFT).
TextFunction Calling在Dify驱动的大模型SaaS平台上开发Agent应用流程，支持行业场景的知识检索与调用组合。通过Workflow将知识库检索、LLM调用、条件分支与结构化输出进行编排，使业务流程可复用且可扩展。该工作为后续对话与任务执行提供可控的自动化推理与功能调用路径。• 基于Dify Workflow编排检索/调用/分支/结构化输出 • 构建可复用的行业Agent业务流程 • 支撑知识库检索在业务中的落地 • 促进Agent在复杂流程中的稳定输出
TextIn this LLM operations internship, you built and maintained an evaluation and optimization workflow for large language models. You analyzed model weaknesses using systematic error categorization and attribution templates. You improved prompt quality and agent execution reliability through iterative tuning and failure diagnosis. • Constructed an LLM evaluation system with Badcase-based structured analysis • Developed an error label taxonomy and root-cause attribution templates • Optimized prompts and Few-shot strategies to improve stability and business controllability • Tuned agent tool-calling and multi-step reasoning to address failures, parameter issues, and context loss.
TextFunction Calling针对智能体工具调用与多步推理链路中的失败与错误进行定位与调优。重点排查工具调用失败、参数生成错误与上下文丢失等关键问题，并推动链路提升可靠性。通过链路级问题分析提升复杂业务任务的执行稳定度。• 诊断工具调用失败原因与失败模式 • 修复参数生成错误与调用接口对齐 • 处理上下文丢失导致的推理中断 • 提升多步推理与工具调用的端到端可靠性
TextPrompt + Response Writing (SFT)在实际业务场景下对Prompt与Few-shot策略进行优化，以减少模型输出偏离与不稳定现象。针对回答偏离、格式不稳定、逻辑跳步等问题迭代提示策略，提高输出一致性与业务可控性。将改进与评测结果相结合，形成可复用的提示优化思路。• 设计并迭代Prompt/Few-shot策略 • 减少回答偏离与格式异常 • 缓解逻辑跳步并提升连贯性 • 将优化效果对齐业务可控性目标
Text负责构建LLM评测与优化体系，产出用于能力短板分析的错误归因与标签材料。围绕指令遵循、上下文理解、逻辑推理、工具调用与检索召回等维度进行系统化Badcase分析。通过整理错误标签体系和归因模板，支撑后续模型迭代定位改进方向。• 建立并维护Badcase系统化分析方法与错误标签 • 覆盖多维度评测场景：指令遵循/上下文/逻辑/工具调用/检索召回 • 沉淀错误归因模板以支撑模型能力分析 • 为Prompt与策略优化提供数据化依据
Master of Science, Information and Communications Engineering
Bachelor of Science, Information and Communications Engineering
Agent Application Development Intern
LLM Operations Intern
