Cambridge Research Intern – AI Code Generation for New Compute Experiences

Developments in generative AI are transforming how we interact with technology including productivity tools such as Microsoft Office. Microsoft Research Cambridge (UK) is seeking motivated research interns with a background in natural language processing, machine learning, or programming languages to join our team of researchers and engineers working to build new compute experiences and applications. 

Our focus is on the development and application of code-generation language models to empower end-users with richer computing capabilities and act on the user’s behalf. 

 
We are Interested in a Broad Range of Areas  
  

• Enhancing code-generating language models through finetuning, in-context learning, and post-training techniques  

• Generating synthetic training and evaluation data for new use-cases  

• Applications of generative AI to data analysis, dynamic interface generation, and productivity tools such Microsoft Excel and Word.  

• Using insights from type systems, static analysis, program synthesis and machine learning to debug and repair generated computations by LLMs   
    

We are a global team with members and collaborators across Microsoft including the ENCODE project, Calc Intelligence team (MSR Cambridge), RiSE (MSR Redmond), Excel, and PROSE. Our team’s research is used by millions of customers, most recently through Microsoft Excel and Excel Copilot. Our work has featured on the Microsoft Wall of Fame, and even XKCD. 

Successful candidates will join us for spring/summer 2025 in Cambridge UK and must be able to start by 15 June 2025. This posting will be active for 30 days, and all candidates applying within this time are considered on an equal basis. 

Required/Minimum Qualifications:   

• Must be enrolled in a PhD program in Machine Learning, Computer Science (or related discipline).   

• Experience in LLM research in one or more of the following areas: code generation, semantic parsing, fine-tuning, reinforcement learning, synthetic data generation, neuro-symbolic programming, benchmarking and evaluation.  

• Experience implementing new prototypes or conducting experiments using software development practices such as version control and code review.  

• Interns are expected to be physically located in Cambridge UK.  

Preferred/Additional Qualifications: 

• Strong communication skills, especially scientific writing and communication.  

• Publications in relevant NLP, ML, PL conferences or journals is preferred (e.g. NeurIPS, ICML, ICLR, ACL, EMNLP, POPL, PLDI, OOPSLA).  
• Ability to work in an interdisciplinary collaborative environment with people from different technical backgrounds.  

• Invent technical solutions to enable new and better user experiences.  

• Develop and evaluate prototypes in collaboration with our research and engineering teams.  

• Write and present your findings in technical documents or research papers.