STMicroelectronicsSenior Design Engineer
Nov. 2020 - May. 2023Project: Verification of Digital Blocks of Analog In-Memory Compute IP for AI/ ML Applications -> Utilized Python and PyTorch to generate vectors for Neural Processor verification, highlighting AI hardware verification skills. -> Performed Quantization-Aware Training to create 4-bit/ 4-bit (Features / Weights) precision models like ResNet-18, VGGNet which can be deployed on In-Memory Compute IP for Low Power AI applications. -> Crafted Python scripts to analyze and optimize power usage via register activity analysis.
Established an SV-UVM based transaction-level verification environment for digital blocks, ensuring scalability and reusability. -> Designed and implemented functional coverage, sequences, and scoreboards for key components, facilitating IP level reuse and aiding in debugging and specification refinement. -> Introduced a self-checking mechanism in the SOC environment, enhancing data verification accuracy. Project: SOC-level verification of Digital In-Memory Compute IP for AI/ ML Applications -> Transitioned the verification environment to the Palladium Z1 Emulation platform, maintaining portability between simulation and emulation. -> Enhanced expertise in AMBA protocols (AXI4, APB, AHB) and developed a System Verilog-based JTAG VIP for SOC stimulus delivery and post-silicon validation. -> Executed comprehensive verification of the DIMC Neural Processor SOC, including test planning, functional testing, and debugging across design hierarchies. -> Conducted power analysis and performance benchmarking for AI models using the emulation platform, optimizing digital NPU efficiency. -> Created a Python-based debugging tool for visualizing AHB/AXI transactions, reducing debug time and improving performance analysis. -> Spearheaded junior engineer teams, mentoring on complex verification projects and developing applications in Computer Vision and NLP for an application-level understanding.