Morteza Rashidinia

Software Engineer Ii @Cerebras

Toronto, ON, CA

Key developer of the inference service at Cerebras, building low-level system software and performance-critical components for LLM workloads on wafer-scale AI hardware.• Designed and implemented optimized compute kernels for KV-cache operations (attention masking, scatter/gather writes, etc.) and high-performance matmuls for transformer inference• Identified performance bottlenecks and drove architectural and kernel-level optimizations to improve token throughput across multiple LLM models under strict memory and hardware constraints• Debugged and resolved kernel-level issues and silicon-level hardware defects directly on-chip

Sharif University of Technology

Bachelor's degree

The University of British Columbia

Master's degree, Computer Engineering