Skip to content
View QuantumDrizzy's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report QuantumDrizzy

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
QuantumDrizzy/README.md

Antonio Zambudio

Research Software Engineer focused on high-performance systems for AI and scientific computing.

I build bare-metal systems that combine GPU-accelerated simulation, quantum & tensor-network computing, and agentic architectures — emphasizing performance, low-level optimization, and sovereign, local-first design. No cloud by default.

Core Focus Areas

  • High-performance systems & GPU acceleration (CUDA, sm_120 Blackwell)
  • Tensor networks & quantum-inspired computing
  • Real-time & embedded systems (Rust)
  • Agentic architectures & computational-neuroscience foundations

Selected Projects

  • KHAOS — Closed-loop BCI kernel with sub-100µs CUDA DSP latency, compile-time-enforced neuroethics, and post-quantum cryptography.
  • SUBSTRATE — Multi-scale physics & quantum-biology simulation framework; hand-written CUDA (sm_120) with an honest roofline (kernel-only 3×→154× vs JAX-CPU) and tensor-network solvers.
  • BLACKWALL — Honest precision-spectrum GEMM roofline on Blackwell (sm_120): FP32 → FP4 measured, FP4 at 20× FP32 via cuBLASLt, anchored to the computed peak.
  • TESSERA — Neural-guided real quantum annealing via tensor networks (MPS/TEBD); a GNN learns the schedule, all on a local GPU.
  • Blaze — Tensor-Train / MPS compression for high-order scientific & quantum-state data, with GPU SVD and an MPS↔circuit bridge.
  • HELIOS — 24/7 predictive DC-microgrid controller — Rust MPPT loop + CNN-LSTM forecasting, post-quantum trust anchors.
  • PROMETHEUS — Distributed LLM pretraining framework — PyTorch FSDP ZeRO-3, activation checkpointing, Flash Attention 2.
  • EIGEN — Quantum-inspired threat modeling via transverse-field Ising models and a pure-Rust Lanczos solver.
  • KINECT-NIR — Real-time IR object detection & tracking — CUDA CA-CFAR kernel + TensorRT INT8 + Kalman tracking.
  • chiron — Bare-metal genomic assembly — lock-free k-mer counting, de Bruijn graphs, and a quantum walk for repeat resolution.

(more: SESHAT — QUBO/simulated-annealing on the undeciphered Linear A script)

Stack

CUDA (sm_120) · C++17 · Rust · Python · PyTorch · cuBLAS/cuBLASLt · TensorRT · Cirq · SLURM

Performant, sovereign systems at the intersection of AI, simulation, and quantum computing — bare-metal, local-first, honest benchmarks.

Pinned Loading

  1. KHAOS KHAOS Public

    Closed-loop BCI kernel with neuroethics enforced at compile time, a sub-100µs CUDA DSP pipeline, and post-quantum sovereignty. Bare-metal, zero-cloud.

    Python

  2. HELIOS HELIOS Public

    24/7 predictive DC-microgrid controller — Rust MPPT loop (100ms tick) + CNN-LSTM irradiance predictor, post-quantum trust anchors, egui dashboard. Bare-metal, no cloud.

    Rust 1

  3. SUBSTRATE SUBSTRATE Public

    Multi-scale electromagnetic computational framework — lattice gauge theory, quantum biology and geomagnetic risk on one GPU. Hand-written CUDA (sm_120) + tensor networks, no cloud.

    Python

  4. BLACKWALL BLACKWALL Public

    Honest precision-spectrum GEMM roofline on NVIDIA Blackwell (sm_120) — FP32 → FP4, measured. FP4 (nvfp4) at 20× FP32 via cuBLASLt, anchored to the computed peak. No inflated numbers.

    Cuda

  5. TESSERA TESSERA Public

    Neural-guided real quantum annealing via tensor networks — a GNN learns the schedule, an MPS engine simulates the transverse-field Ising adiabatic process on your own GPU. Local, sovereign, honest …

    C++

  6. Blaze Blaze Public

    Tensor-Train / MPS compression for high-order scientific & quantum-state data — a specialist (not general) compressor with an honest compressibility diagnostic, GPU SVD, and an MPS↔circuit bridge.

    Python