Hi! I'm Raja...

I'm a Research Fellow (Pre-doctoral Researcher) in the AI-infrastructure team at Microsoft Research India (MSR-I). In 2023, I graduated from the Undergraduate Programmes at the Indian Institute of Technology, Bombay, where I earned my B.Tech (with Honors) in Computer Science. My primary interests lie in Systems for ML, Compute Express Link (CXL), Networking, and Systems in general.

At MSR-I, I am working with Dr. Nipun Kwatra and Dr. Ramchandran Ramjee on improving GPU utilization for LLM inference. Specifically, our focus is on the communication aspect of multi-GPU LLM inference, which currently lies in the critical path and impacts both latency and efficiency. We are looking at approaches to mitigate these overheads by hiding them behind existing computations, particularly for Mixture of Expert (MoE) models as most large high-accuracy models are MoEs. During my undergrad, I worked with Prof. Purushottam (Puru) Kulkarni on CXL and persistent memory.

Besides research, I have also been curious about other sides of the classroom. To explore it, I served as an Undergraduate Teaching Assistant for several CS courses, including Computer Networks (with Prof. Bhaskar Raman), Operating Systems (with Profs. Puru Kulkarni and Umesh Bellur), and Computer Systems Bootcamp (with Profs. Mythili Vutukuru and Puru Kulkarni).

When the deadlines are quiet, and time slows, you may find me taking an infinite walk or reading a book (mostly Hindi or Urdu poetry). Sometimes, as the sun sets in the western sky, I steal moments from the treasure of time to weave my thoughts into words (check the poetry below!). I am no great poet now, but someday in the future, I hope to be a decent one.

• Research
• Curriculum Vitae
• Poetry
• Notes