- Build code datasets for AI-powered code intelligence
- Develop static/dynamic analysis tools
- Implement real-time code analysis in GitHub Copilot workflows
- Apply code analysis techniques to mission-critical distributed systems, improving reliability and security
- Innovate in code-CoT prompting and programming instruction fine-tuning
- Advance code representation learning with program synthesis techniques
- Research AI-enhanced software analysis applications
- Publish breakthroughs at PLDI/ICSE/FSE conferences
- Strong programming skills in Python/Java/C++ with Linux experience
- Experience in either:
-
- Software analysis tools (AST/CFG analysis) and distributed systems
-
- LLM frameworks (PyTorch/DeepSpeed) and prompt engineering
- Understanding of compiler principles or code intelligence paradigms
- Publications in software engineering/PL conferences preferred
- Must be available for a full-time internship for 6 months or more.
- 构建支持代码智能的大模型数据集
- 开发动态/静态分析优化工具
- 在GitHub Copilot中实现实时代码分析
- 将代码分析技术应用于核心分布式系统,提升可靠性与安全性
- 研发代码思维链提示与编程指令微调技术
- 推进代码表征学习与程序合成方法
- 探索AI增强的软件分析技术及应用
- 在PLDI/ICSE等顶会发表创新成果
- 扎实的Python/Java/C++能力,具备Linux开发经验
- 满足以下任一领域经验:
-
- 软件分析工具(AST/控制流分析)与分布式系统
-
- 大模型框架(PyTorch/DeepSpeed)与Prompt工程
- 理解编译原理或代码智能范式
- 在软件工程/编程语言顶会有论文发表者优先
- 每周全职工作5天,能够保证实习6个月以上
Microsoft is an equal opportunity employer. All qualified applicants will receive consideration for employment without regard to age, ancestry, citizenship, color, family or medical care leave, gender identity or expression, genetic information, immigration status, marital status, medical condition, national origin, physical or mental disability, political affiliation, protected veteran or military status, race, ethnicity, religion, sex (including pregnancy), sexual orientation, or any other characteristic protected by applicable local laws, regulations and ordinances. If you need assistance and/or a reasonable accommodation due to a disability during the application process, read more about requesting accommodations.