Mustafa Ali
M.S. CS student and software engineer focusing on AI and systems
Summary
Researcher in compute-in-memory and hardware-aware ML: Mustafa has co-authored multiple peer-reviewed and preprint publications on in-memory computing (SRAM/DRAM), resistive crossbars, and compute-in-memory architectures, contributing technical work on primitives and their impact on ML workloads. arxiv+2
Software engineer focused on scalable systems and AI: He has 3+ years of industry experience building backend infrastructure, microservices, ETL pipelines, and LLM/RAG-style systems, and has shipped products and automation at Atica Global. github+1
Graduate student and GenAI practitioner: Pursuing an M.S. in Computer Science at UMass Amherst while contributing to the Center for Data Science & AI GenAI activities and working as a grader/teaching assistant role. github+1
Creative practitioner and community organizer: Active in film and media on campus (FilmFest coordinator and film/media secretary roles) and maintains a filmmaking portfolio alongside technical work. github+1
Work
Education
Projects
Writing
IMAC: In-Memory Multi-Bit Multiplication and ACcumulation in 6T SRAM Array
January 1, 2020Paper presenting an in-memory multiplication and accumulation mechanism inside 6T SRAM arrays enabling parallel dot-products without bitcell changes; evaluated on LeNet-5 and VGG with reported accuracy and energy/delay benefits versus von Neumann baselines.
Resistive Crossbars as Approximate Hardware Building Blocks for Machine Learning: Opportunities and Challenges
January 1, 2020Co-authored overview on resistive crossbar architectures for in-memory computing, discussing device/circuit non-idealities and architectural/software considerations for ML acceleration.
In-memory low-cost bit-serial addition using commodity DRAM technology
January 1, 2019Work on in-memory addition primitives in commodity DRAM to enable efficient parallel operations for data-intensive workloads (listed among authored publications).