• Master's project at IIT Bombay under supervision of Prof. Bipin Rajendran. (Thesis)

• Received Best MTech Project Award 2015 from Department of Electrical Engineering, IIT Bombay for it.

• Description:

  • Developed a novel supervised learning algorithm named NormAD for training spiking neurons.

  • Demonstrated at least 10X faster convergence with it compared to previous state-of-the-art (Publication).

  • Developed multilayer NormAD - learning algorithm for training multilayer spiking neural networks (Publication). It is spatio-temporal equivalent of error backpropagation suitable for spiking neural nets.

  • Proposed a non-convex optimization method named Normalized Approximate Descent using which all above learning algorithms were derived.

• Bachelor's project at VNIT Nagpur under joint supervision of Prof. R B Deshmukh and Prof. R M Patrikar.

• Received ISSS Undergraduate Award-2013 (2nd place) from Institute of Smart Structures and Systems (ISSS), India.

• Presented paper titled Electrowetting-on-Dielectric based Digital Microuidics: Design, Fabrication and Testing of a low cost test-chip at ISSS National Conference-2013. (PDF)

• Description:

  • Simulated basic droplet handling operations by modeling electrowetting-on-dielectric (EWOD) phenomenon in Comsol Multiphysics.

  • Designed and fabricated a digital microuidics test-chip; then successfully tested droplet transport and merging operations on it.

  • Proposed a hypothesis that in a DMF chip optimal droplet actuation voltage depends solely on the weakest dielectric material in the stack.

  • Proposed a design of a programmable digital microuidics chip.

  • Check the demo videos below.

• Presently working on applications of Machine learning to 5G NR modem design at Qualcomm Wireless R&D.

  • Problem formulation

  • Feature selection and preprocessing

  • Simulation setup for training data generation and/or end-to-end validation

  • Neural network architecture optimization

  • Inference and post-processing

  • integration and end-to-end validation

• Worked on downlink rate adaptation through channel state information feedback for MIMO LTE modems at Qualcomm Wireless R&D Bengaluru.

• Geometric acoustic simulation using ray tracing on NVIDIA GPUs to render 3D audio for a given geometry in VR applications at NVIDIA Pune Design Center.

• Optimizing DSP kernels for SIMD and VLIW based DSP coprocessors at Tensilica Pune (now Cadence).

Worked as a Teaching Assistant in Electrical Engineering Department, IIT Bombay during my masters for following courses/labs.

It was a joy instructing and interacting with conceivably some of the smartest undergraduate students.

Courses/Labs:

• Communication Lab (Autumn 2013 and Autumn 2014),

• Digital Signal Processing Lab (Spring 2014) ,

• Wireless and Mobile Communication (Spring 2015).

Mentored students for lab assignments on analog and digital communication and implementation of primary digital signal processing operations such as FIR and IIR filtering, implementation of FFT, etc. The experiments involved working with DSPs and USRPs. Also made a major contribution to defining relevant course projects for the students.