Resume — Mitchell Bizzigotti

Education

University of California San Diego

Graduate Researcher, University of California San Diego

Jan 2026 – Present

Developed a new LLVM backend to generate code for a modified RISC-V architecture, enabling automatic code generation previously not possible.
Created testing infrastructure to test and verify compiler correctness, helping catch several internal compiler bugs.

Embedded Systems Engineer, GelTech Labs, San Diego, CA

Oct 2026 – Mar 2026

Developed firmware in Embedded C and ARM Assembly to control a variaty of motors and sensors.
Used Python to write scripts that tested device functionality, enabling automated device testing.
Used STM32 Toolchain and Debugger to catch firmware issues that resulted in improved operational reliability.
Collaborated with team of engineers on RCA to help diagnose and solve multiple device problems.
Rewrote GUI application from C# to C++ to enable more reliable firmware integration, better user experience, and faster development.

Teaching Assistant, University of California San Diego

Jan 2026 – Mar 2026

Developed new starter code with Vulkan and C++ to be more up to date with modern C++ software practices.
Organized and conducted discussion sections to support student's learning.

FPGA Guitar Pedal

Worked in a team of 4 using an AMD PYNQ FPGA to create a working guitar pedal.
Modified device drivers to support realtime data transfer between FPGA and ARM subsystem using an AXI-lite interface

Pipelined Out-of-Order Processor

Worked in a team of 2 to design, test, and implement a modern MIPS processor using SystemVerilog in a Linux environment with Verilator.
Implemented out-of-order execution with register renaming to increase processor performance by up to 50%.
Designed a custom non-blocking data cache to increase memory throughput by 1%

Developed a dynamic multi-threaded chunk remeshing system with Vulkan to optimize terrain updates without blocking the render thread
Implemented level-of-detail voxel streaming to allow more than 100× the draw distances.
Integrated greedy meshing algorithm to decrease cost of CPU to GPU data transfer.

Designed an open-source performance-oriented scripting language in C with zero dependencies to target real-time systems.
Built a custom bytecode interpreter to support compile-time code execution and enable advanced metaprogramming features.
Utilized arena allocation strategy to decrease total compilation speed, resulting in a 2x speed up in parsing.

Implemented particle-based physics models with neighbor searching and Eulerian integration techniques to simulate real fluid behavior.
Leveraged modern GPU compute capabilities through WebGPU compute shaders to simulate and render thousands of fluid particles in real-time.
Optimized simulation performance by minimizing unnecessary memory transfers to achieve over 60 frames per second with tens of thousands of particles.