Summary
Jennifer Zhou is a graduate student researcher at Harvard specializing in memory systems, HW/SW co-design, and characterization of LLM inference workloads. With six years of experience spanning tapeouts, GPU/DRAM simulator integration, and custom memory-controller design for TSMC16, she has boosted ML streaming bandwidth by 1.7x and led ML-guided DSE of chipletized architectures. Her work models real-world LLM request patterns to quantify throughput and latency trade-offs across disaggregated inference deployments, bridging rigorous academic research with practical system design. A UC Berkeley EECS alum who helped tape out prototype cores and taught VLSI and FPGA labs, she combines deep hardware expertise with software tooling experience from an internship on Facebook’s Presto team. Based in Cambridge, MA, she is equally comfortable authoring simulators and leading teaching teams, and she leverages fine-grained design tuning to translate workload insights into silicon-level optimizations.
6 years of coding experience
2 years of employment as a software developer
Bachelor of Science - BS, Electrical Engineering and Computer Science, Bachelor of Science - BS, Electrical Engineering and Computer Science at UC Berkeley College of Engineering
Graduate Student, Computer Science, Graduate Student, Computer Science at Harvard University
High School Diploma, High School Diploma at BASIS Independent Silicon Valley
Chinese, English