Visiting Research Assistant, Johns Hopkins University (Fall 2023 - Present)

• • Developed adaptive multimodal DNN compression techniques, leveraging novel mixed precision quantization, cyclic sparsification and knowledge distillation, achieving up to 200× model size reduction while retaining over 95% of the original performance metrics for diverse vision and audio processing tasks.

  • Deployed compressed multimodal DNN models across a range of resource-constrained edge devices, enabling efficient real-time applications.

  • Actively contributed to drafting grant proposals for major funding agencies including NSF, DARPA, and ARL, leveraging in-depth research insights and strategic planning.

Embedded AI Application Intern, Analog Devices Inc. (Summer 2023)

• • Assembled an audio-visual hand gesture recognition dataset by synthesizing various open-source datasets, tailored for ADI’s internal use, enhancing multimodal AI application deployment capabilities.

  • Developed and benchmarked different CNN and TCN multimodal models, employing diverse fusion techniques for hand gesture recognition, utilizing PyTorch based AI8x modules; achieved 5% performance boost in the hybrid fusion model, optimized for deployment on ADI’s MAX78000/MAX78002 hardware.

Embedded Machine Learning Intern. Starkey Hearing Technologies (Summer 2022)

• • Developed refined DTLN and TASNet-based real-time speech enhancement algorithms, achieving 97% performance retention with efficient compression.

  • Deployed speech enhancement algorithms on Raspberry Pi 3B with an end-to-end delay of only 4ms, ensuring adherence to real-time implementation constraints.

Research Intern, Nokia Bell Labs                                       (Summer 2021)

• • Developed a mixed precision quantization technique for ResNet models, yielding a 6.7×size reduction and 3.6×speed increase, with 98% performance retention.

  • Optimized multiple ResNet models running on shared hardware, ensuring efficient concurrent operation with balanced latency-accuracy trade-offs.

Research Intern, Army Research Laboratory (ARL) (Summer 2020)

• • Investigated reinforcement learning (RL) in the Starcraft-II environment.

  • Applied imitation learning to develop an algorithm that mimics expert Starcraft-II tactics.

Graduate Research Assistant, Energy Efficient High Performance Computing (EEHPC) Lab, UMBC (Summer 2019 - Spring 2023)

• • Led the UMB-ICTR funded Covid-Matter project, collaborating with physicians to develop scalable multimodal sensory ML algorithms for detecting the severity of respiratory diseases, resulting in published research findings .

  • Developed an energy efficient and flexible multichannel electroencephalogram (EEG) artifact detection and identification techniques using depthwise separable CNN and LSTM, funded by NSF and ARL.

Graduate Teaching Assistant, Department of Computer Science and Electrical Engineering, UMBC (Fall 2018 - Spring 2019)

• • Courses: Principle of VLSI design, Embedded System and C programming, Principle of Digital Design.