PhD & Postdoc

Ramesh Subramaniam

Electrical Engineering · IIT Madras

Ramesh Subramaniam is a PhD candidate in Electrical Engineering at IIT Madras, supervised by Prof. Kamakoti Veezhinathan. His thesis designs analogue in-memory compute units and spike-coding circuits for neuromorphic AI inference acceleration, demonstrating three-orders-of-magnitude energy reduction versus GPU baselines on embedded vision tasks. Ten publications in IEEE Transactions on Very Large Scale Integration and DATE conference proceedings document his chip architectures. Ramesh taped out two test chips through IMEC and received a Qualcomm Innovation Fellowship. He has accepted a principal engineer position at Intel Labs Bangalore to develop neuromorphic processing units for autonomous vehicle sensor fusion. His neuromorphic inference unit achieves 50 TOPS per watt on ResNet-50 benchmarks, enabling always-on vision AI in battery-constrained edge devices.

Thesis Topic

Neuromorphic analogue in-memory compute units for edge AI inference acceleration

Transition Signals