GRD-Net
Generative-reconstructive-discriminative anomaly detection with region-of-interest attention.
Projects
Some projects are published and stable; others are active research or private work summarized at a safe level. The common thread is practical engineering around machine learning systems.
Generative-reconstructive-discriminative anomaly detection with region-of-interest attention.
Integration of deep generative anomaly detection into a high-speed industrial line.
Covariance-aware and streaming-compatible PatchCore research for industrial anomaly detection.
Co-authored language model architecture research around graph-structured memory.
Local-first agent runtime and mini RAG experiment that may later power the personal assistant.
Governance and control-plane system for coding-agent workflows.
Small benchmarking and stress-testing tool for CPU, CUDA, and MPS tensor operations.
Portable spotlight with PCB assets, ATtiny1616 firmware, sensors, PWM, buttons, and OLED status.
Unreal Engine C++ event bus project compatible with modern UE 5.x versions.
Simple C++ Drone Companion Unreal Engine 5.5 project.
Older discontinued GRD-Net implementation kept as historical and reproducibility context.
Upcoming edge deployment line for MH-PatchCore distillation and LuckFox Pico Mini B firmware.
Portfolio RAG assistant backend for recruiter-facing questions about my public profile.