I build production-grade tools and distributed systems across the stack — from low-level Java networking and event-driven backend architecture to TypeScript frontends and mobile applications. Strong interest in open source, performance engineering, and developer tooling.
Currently building at @ordnary-com.
Enterprise-grade shared MongoDB connection layer for distributed server networks ·
Java·MongoDB·Caffeine·REST
A production-ready data synchronization solution architected for high-concurrency server networks. Instead of each plugin managing its own database connections — causing connection storms at scale — NetworkDataAPI exposes a shared MongoDB connection pool consumed by all plugins simultaneously, reducing total connections by 80%+ on a typical 5-server network.
Architecture highlights:
- Multi-module Maven project with isolated
core,paper,bungee, andexample-pluginmodules — the core is platform-agnostic, platform modules provide thin adapter layers - Caffeine-backed in-memory cache with configurable TTL and max-size, achieving 85–95% cache hit rates in production workloads
- Fully async API built on
CompletableFuturethread pools (core-pool-size/max-pool-sizeconfigurable) — zero blocking on the main server thread - Optional REST API (Spark Java) exposing
/api/player/{uuid}CRUD endpoints, secured viaX-API-Keyheader, for external service integrations - CodeQL, Maven CI, and build/release pipelines via GitHub Actions
- Benchmarks:
<5msfor cached reads,<50msfor live MongoDB queries, supports1000+concurrent operations
🖼️ YEETIFF
Custom image container format with transcoding pipeline ·
Python,Rust
An experimental image file format specification and transcoding toolchain — Yet Even Extremely Expressier Transcoded Image File Format. Built to explore custom binary encoding, metadata schemas, and image pipeline design at a low level.
Open source thrives on collaboration. Feedback, bug reports, and pull requests are always welcome.
Built with care · @7txr



