About this role
Job Description:
- 负责 AI Agent 系统的设计、开发与迭代，涵盖 规划推理、工具调用、多轮对话、记忆管理 等核心能力建设；
- 基于大语言模型（LLM）构建 Agent 编排框架，实现 任务分解、ReAct/Plan-and-Execute 推理、多 Agent 协作 等高级编排策略；
- 开发与维护 Agent 的 工具生态 （Function Calling / Tool Use），对接内部业务系统、数据库、搜索、代码沙箱等外部能力；
- 设计并实现 长期记忆与上下文管理 方案（向量检索 / RAG / KV Memory / 会话状态持久化），持续优化长程任务表现；
- 参与 Agent 的 评测与对齐 工作：构建评测数据集、设计自动化评测 Pipeline、分析 Bad Case 并迭代优化；
- 与算法团队协作，推进 模型微调（SFT / RLHF） 在 Agent 场景的落地，提升工具调用准确率与推理链质量；
- 编写技术设计文档，参与 Code Review，保障代码质量与系统可维护性；
- 关注行业前沿（Multi-Agent、Computer Use、World Model 等），定期进行技术调研与原型验证。
At DXC Technology, we believe strong connections and community are key to our success. Our work model prioritizes in-person collaboration while offering flexibility to support wellbeing, productivity, individual work styles, and life circumstances. We’re committed to fostering an inclusive environment where everyone can thrive.
Recruitment fraud is a scheme in which fictitious job opportunities are offered to job seekers typically through online services, such as false websites, or through unsolicited emails claiming to be from the company. These emails may request recipients to provide personal information or to make payments as part of their illegitimate recruiting process. DXC does not make offers of employment via social media networks and DXC never asks for any money or payments from applicants at any point in the recruitment process, nor ask a job seeker to purchase IT or other equipment on our behalf. More information on employment scams is available here .