Summary
Joshua Bakita is a research-focused software engineer and PhD candidate with 13 years of systems and real-time software experience, specializing in making GPUs fast and predictable for autonomous systems. Based at UNC Chapel Hill, he builds low-overhead GPU partitioning, reverse-engineering, and memory-oversubscription tools that have driven peer-reviewed breakthroughs and over 13,000 downloads. His work has demonstrated substantial runtime gains (e.g., 41% lower mean and 52% lower max execution times vs. MPS) and revealed unsafe assumptions in prior real-time GPU analyses. A frequent Linux kernel contributor and LITMUS-RT maintainer, he also brings practical industry experience from Waymo, GM, and Microsoft where he optimized GPU and browser rendering pipelines. He combines deep OS and concurrency insight—NUMA-based O(1) page coloring and SMT-aware schedulers—with hands-on CUDA/C++ implementation. Outside core research, he’s taught systems programming at scale and ported tracing tools to ARM, showing a knack for translating theory into deployable systems.
13 years of coding experience
3 years of employment as a software developer
High School, High School at Homeschool
Doctor of Philosophy - PhD Computer Science Real-Time Systems, Doctor of Philosophy - PhD Computer Science Real-Time Systems at University of North Carolina at Chapel Hill