Mohammed Muddasser

Virtual Prototyping (SytemC) and ECU Virtualization solutions for pre-silicon and post-silicon use cases of automotive embedded software.

C++ based embedded software development for Adaptive AUTOSAR Diagnostics Management, deployed in Linux and QNX environments. Working on Diagnostics over Internet Protocol (DoIP) module within the cluster.

Commissioning of Livox LiDAR devices. LiDAR data processing, tooling solutions and image processing.

Commissioning of NVIDIA Jetson devices for image processing algorithms with GPU acceleration.

Validation, documentation and delivery of virtual platforms as per the customer requirements. Project build solutions for different IDE's. Providing necessary tooling automation and testing solutions.

High Performance Programming with Graphic Cards – GPU based acceleration of sparse matrix computations for learning based image super resolution: A computed sparse matrix is applied to downsample a high resolution (HR) image to low resolution (LR) image for feature extraction and joint learning from both the HR and LR images. The CPU based matrix computation algorithms are accelerated with newly designed GPU based parallelism. These computations are accelerated on AMD GPU's using OpenCL.

2. Test Framework Development: • Integration Test Framework: Feature development to support testing of automotive Ethernet in an existing perl based environment. Automation with Vector Canoe, CAPL and COM interfacing. • Basic Software Simulation: Development of a plug-in component over an existing C++ based simulation environment to support the testing of diagnostics over Ethernet (IP). 3. Continuous Integration and Testing: Interoperability check between modules in a development stream.

- Estimating confounding physical quantities (mass, friction, etc.) of stacked cubes. Investigative study of published work consisting of a seq2seq model with Graph Convolutions and Gated Recurrent Units. Estimating the center of mass of the stack using confounders. - Comparative study of stack stability prediction (from images), using the learnt confounders with previous best performing method which used an Inceptionv4 model. - Improving the model performance additional perception inputs.

- High Performance Computing with GPU's - Fachpraktikum - Detection and Pattern Recognition, Deep Learning, Advanced mathematics - Embedded Systems, Real Time Programming, Data Structures and Algorithms - Communications, Mobile Computing - Digital System Design, Computer Architecture and Org

Electronics and Communications