My Work
Projects, code, and experiments I've been building
Active Projects
🎮 Pokémon Red Deep RL System
End-to-end AI training pipeline for autonomously learning and playing Pokémon Red. Custom Game Boy emulator (~3,200 FPS) + Stable Baselines 3 PPO with 64 parallel training environments via CUDA.
Stack: Python, PyTorch, CUDA, Stable Baselines 3, Game Boy Emulation
🕹️ rlgb Game Boy Emulator
Cycle-accurate Game Boy emulator written in C. Full SM83 CPU implementation, scanline-accurate PPU, MBC support, and high-performance runtime (~3,200 FPS). Used as the core for RL training and experimentation.
Stack: C, Assembly, High-Performance Computing
View on GitHub →⛏️ mc-rust Minecraft Server
Minecraft server implementation in Rust using the Valence framework. Full protocol support, chunk loading, entity systems, and custom game mechanics for Skyblock-style gameplay.
Stack: Rust, Valence Framework, Async Networking
🌐 BuiltNotTaught.com
This website itself! Built with Rust, Axum web framework, SQLite database, and modern web practices. Demonstrates full-stack development with async Tokio runtime and Stripe payment integration.
Stack: Rust, Axum, SQLite, Tera Templates, Stripe API
View on GitHub →🔧 Systems & Infrastructure
Various infrastructure and systems projects including home network setup (pfSense), server administration, DevOps practices, and Docker containerization experiments.
Stack: Linux, pfSense, Docker, Kubernetes, Nginx, systemd
⚡ CUDA & GPU Computing
GPU-accelerated computation experiments including CUDA kernel development, parallel processing optimization, and high-performance computing for ML training pipelines.
Stack: CUDA, cuDNN, PyTorch, GPU Optimization
Archive & Experiments
Beyond active projects, I maintain an archive of experiments, iterations, and learning materials. These include:
- Pokemon RL Experiments - Multiple iterations of the RL training pipeline with different architectures and reward shaping strategies
- Emulator Variants - Various Game Boy emulator implementations (C-based, Rust-based, vectorized versions)
- Training Runs & Checkpoints - Saved training sessions, model checkpoints, and TensorBoard logs for reproducibility
- System Snapshots - Archived configurations of development environments and infrastructure setups
- Research & Documentation - Write-ups on emulation, RL training, performance optimization, and deployment practices
All code follows principles of clean architecture, documentation, and reproducibility. Projects are continuously iterated on based on learnings and new ideas.