Muhammad Mohsin Ghaffar

Design end-to-end DNN development pipeline including data curation, model adaptation, and optimizations for classification, object detection, and segmentation tasks. Optimize DNN deployment for edge platforms for real-time inference and energy efficiency across diverse edge platforms (MCUs, CPUs, GPUs, FPGAs, and NPUs). Develop and verify custom FPGA/NPU accelerators, leveraging dataflow and systolic array architectures to boost performance and reduce power consumption. Project management and reporting.

Design end-to-end DNN development pipeline including data curation, model adaptation, and optimizations for classification, object detection, and segmentation tasks. Optimize DNN deployment for edge platforms for real-time inference and energy efficiency across diverse edge platforms (MCUs, CPUs, GPUs, FPGAs, and NPUs). Develop and verify custom FPGA/NPU accelerators, leveraging dataflow and systolic array architectures to boost performance and reduce power consumption. Project management and reporting.

Developed FINN-L framework for deploying LSTM models on FPGA.

Led a research group focused on AI model development, optimization, and deployment. Prepared funding proposals and ensured the timely completion of project milestones.

Implemented and validated the DRAM and VGA controller for Xilinx FPGAs. Implemented an I2C driver for the STM32F4 to enable communication between multiple sensors.

Design and Implementation of a Multi-Port Multi-Channel 3D-DRAM Controller Frontend Complete verification of the design with the memory controller backend and 3D-DRAM Performing ASIC synthesis and generating area, power and performance estimations

Testen und Inbetriebnahme of Optical Fiber Geräte. Installation, Wartungen und Kundendienst dieser Geräte.

Eingebettet Systeme, Digital Design, Hardware Entwicklung, RTOS, Verification of Digital Systems

Electronics, Analog and Digital Electronics, Control Systems, Power Electronics, Digital Signal Processing, Microwave Theory