Hello, I’m Abiel. I am an undergraduate computer engineer at FIU.

My primary focus lies in hardware-aware programming to accelerate machine learning inference. Often utilizing CUDA, C++, and PyTorch alongside profiling tools to guide my optimizations.

I am motivated by the human experience my optimizations bring to the end-user and the real-time feedback loop of collaborating on unsolved problems, especially in real-time systems for embodied AI.

Background

Previously interned at NVIDIA on the ChipNemo team. I also worked at FIU's Applied Research Center under Dr. Himanshu Upadhyay. In both roles, I built LLM serving, retrieval, and deployment infrastructure and experimented with compiler front-ends for domain-specific languages.

Activity

  • Currently: Working on flash-recon, real-time monocular SLAM with custom fused CUDA kernels, Gaussian Splatting, and DepthAnythingV2 on a single GPU. link
  • Feb 2026: Built pperf, a hierarchical profiler for hunting bottlenecks in complex GPU systems. link
  • Feb 2026: Published a technical deep dive showing how I reduced computer vision pipeline latency from 22ms to 11.3ms with microsecond consistency. link
  • Nov 2025: Excited to share that I will be joining Apple AIML this summer as a Machine Learning Intern, possibly working on efficient on-device computer vision!