NEW DELHI: Indraprastha Institute of Information Technology (IIIT) Delhi and the Indian affiliate of Marvell Technology, Inc., have launched a new programme – Networks for AI/ML Systems.
The new programme is an advanced academic-industry course focused on the systems layer of modern AI, where networking fabrics, memory hierarchies, and distributed execution determine performance at scale, an official statement released in this regard said.
The programme was jointly planned to educate students on the realities as they appear in production AI infrastructure.
The course was co-designed and is co-taught by Dr Rinku Shah (IIIT-Delhi) and Abed Mohammad Kamaluddin, director at Marvell, with ongoing technical mentoring from Marvell engineers and architects.
The IIIT Delhi and Marvell are collaborating on research in AI networking and systems, with the course reinforcing and feeding into these joint research directions.
“Networks for AI/ML Systems is India’s first and the first known course of its kind globally. By integrating AI networking, CXL-based memory systems and AI-scale simulation, the course exposes students to system-level challenges rarely addressed in traditional curricula,” said Prof. Pushpendra Singh, Head, Department of Computer Science and Engineering, IIIT-Delhi.
Nearly 45 students from BTech, MTech and PhD programmes participated in the initial course, which followed a project-first, peer-driven pedagogy.
A post-course survey showed a clear increase in student confidence in reasoning about AI systems, fabrics and distributed machine-learning workflows.
Student projects will span AI fabrics and transports, CXL-aware training and inference, distributed training and serving pipelines, in-network compute, telemetry-driven optimization and ML-based security and observability. Students consistently identified the open-ended, systems-heavy nature of the projects as the most influential aspect of their learning, enabling them to explain complex system behavior, rather than just implementing tools.