Overview

‍‍

Start Date:  March 2026
Duration: 12 weeks
Location: Cambridge, UK

‍

At MSR Cambridge we are creating the future of AI infrastructure by tackling ambitious challenges and developing disruptive technologies for Microsoft's Cloud and AI platforms.

‍

AI workloads are growing at an unprecedented pace and are currently the most pressing challenge in modern computing. They require systems at unprecedented scale to sustain these demands. This massive and growing demand is hitting the performance, power and energy limitations of today’s hardware, creating a perfect storm of challenges.

‍

Our multidisciplinary research team spans computer systems, AI, and hardware, allowing us to take a full-stack approach to these challenges: from the models to the runtimes to the hardware. We work closely with product teams across the company to understand real AI workloads at scale as well as new and emerging hardware technologies, and partner across industry and academia to drive changes across the technology ecosystem.

‍

‍

We have three research efforts, each of which has multiple potential intern projects:

• Networking for AI: more efficient, circuit-switched network architectures for AI collective communications, and an AI-first network stack.

• Runtimes for next-generation heterogeneous infrastructures, in which GPUs co-exist with emerging technologies such as near-memory compute and new classes of memory (e.g., HBF, MRM).

• Memory subsystem co-design: co-designing AI workloads with the memory subsystem, to achieve significant performance gains and energy savings. This includes the memory chip internals, the memory controller, and new memory cells.  

‍

The internship will require using a range of techniques from roofline modelling, simulation tools (ns3, Cacti-3D), to testbed implementations (NCCL, etc.). The exact mix will depend on the project needs, as well as on your skills and experience.  

If you want to have real-world impact, have a taste for designing and evaluating systems at scale, and expand your technical skills, this internship is for you.  

‍

‍

Here are a few recent publications and talks from our team that give a flavour of the kind of research that we do:

Storage Class Memory is Dead, All Hail Managed-Retention Memory: Rethinking Memory for the AI Era (HotOS ’25)

Good things come in small packages: Should we build AI clusters with Lite-GPUs? (HotOS ’25)

Towards Memory Specialization: A Case for Long-Term and Short-Term RAM | (DIMES ’ 25)

Analog optical computer for AI inference and combinatorial optimization (Nature)

Mosaic: Breaking the Optics versus Copper Trade-off with a Wide-and-Slow Architecture and MicroLED (SIGCOMM 2025 Best Paper)  

Project Silica: Towards Sustainable Cloud Archival Storage in Glas (SOSP 2023)

Systems Research in the AI Era (APSys 2025 keynote)

‍

‍

‍Responsibilities‍

Research Interns put inquiry and theory into practice. Alongside fellow doctoral candidates and some of the world’s best researchers, Research Interns learn, collaborate, and network for life. Research Interns not only advance their own careers, but they also contribute to exciting research and development strides. During the 12-week internship, Research Interns are paired with mentors and expected to collaborate with other Research Interns and researchers, present findings, and contribute to the vibrant life of the community. Research internships are available in all areas of research, and are offered year-round, though they typically begin in the summer.

‍

‍

‍Qualifications‍

Required/Minimum Qualifications:    

• Currently enrolled in or accepted to a PhD program in Computer Science, Machine Learning, Electrical Engineering or a related STEM field; or proven comparable experience in industry.

‍

Preferred/Additional Qualifications:

• Experience with LLM architectures and distributed inference/training
• Understanding of AI hardware architectures, e.g. GPUs
• Experience with low-level AI software stack such as CUDA and NCCL.  
• Experience with network simulation (e.g. ns3)  
• Experience with hardware simulation (e.g. Cacti-3D)
• Experience with analytical performance modelling (e.g. alpha-beta model)
• Be able to work in a cross-functional and multi-disciplinary setting.  
• Proficient software development skills, preferably in C++ and Python.

Candidates with one or more of the above qualifications are encouraged to apply.

‍