fsdfas32/ejrniiii8-cloud

KR 한국어	US English	JP 日本語	CN 简体中文	ES Español
FR Français	DE Deutsch	RU Русский	SA العربية	IN हिन्दी

---

Julian Vance

Systems & Storage Engine Engineer • Distributed Systems Architect • Open-Source Tinkerer

Hi there! I am Julian, a systems engineer obsessed with low-latency storage engines, deterministic consensus models, and low-level Linux performance optimization. Over the past decade, I have been designing infrastructure that operates reliably when the network fails and disk queues back up.

Outside of distributed primitives, I spend my weekends soldering Eurorack synthesizer modules, studying cold-war era telecommunications hardware, and restoring vintage mechanical keyboards.

---

🔭 Active Work & Open-Source Projects

Project	Role / Scope	Primary Stack	Highlights
keg-db	Author & Maintainer	Rust, io_uring, SIMD	Embedded, crash-resilient LSM-tree storage engine tailored for NVMe drives with lock-free memtables and zero-copy block compaction.
wirefall	Lead Contributor	C, eBPF, Go, XDP	In-kernel high-throughput packet telemetry and rate-limiting pipeline running sub-microsecond filtering on 100GbE edge links.
raft-sim	Author	Rust, TLA+	Deterministic discrete-event simulation testbed for Raft consensus under asymmetric network partitions and jitter.
eurorack-dsp	Hardware & Firmware	C++20, ARM Cortex-M7	Open-hardware Eurorack macro-oscillator module featuring bandlimited wavetable and FM synthesis with <1.2ms latency.

---

🛠 Technical Stack & Tooling

Core Systems Programming

• Languages: Rust (1.80+), Modern C++ (C++20/23), Go (1.23+), ANSI C (C99/C11), Zig
• OS Internals: Linux kernel (6.x), eBPF / BCC / libbpf, io_uring, POSIX systems programming, memory mapping (mmap), hugepages

Storage & Concurrency

• Engines & Formats: RocksDB, SQLite internals, Parquet, WAL (Write-Ahead Logging), LSM-Trees, B-epsilon trees
• Concurrency Primitives: Lock-free data structures (crossbeam/epoch-based reclamation), atomics, RCU (Read-Copy-Update), actor models

Distributed Infrastructure

• Protocols & RPC: Raft consensus, Paxos, gRPC, FlatBuffers, Protocol Buffers, QUIC
• Observability & Perf: Linux perf, eBPF traces, Flamegraphs, Valgrind (Callgrind/Helgrind), Cachegrind, Wireshark

Developer Environment

• Platform: Arch Linux (custom minimal kernel) / FreeBSD on workstation; Debian LTS on servers
• Editor & Tools: Neovim (Lua-configured), Tmux, Git, GNU Make / Ninja, CMake, Cargo

---

📐 Engineering Principles

1. Mechanical Sympathy First
   • Software must respect the underlying silicon: CPU cache-line alignment, branch prediction predictability, memory strides, and NVMe page boundaries matter more than premature syntactic elegance.
2. Determinism over Cleverness
   • Distributed algorithms and storage engines must be deterministic and verifiable through property testing and reproducible discrete-event simulation before hitting production.
3. Crash-Only Software (Safe by Design)
   • Systems should recover as safely and swiftly from an abrupt power loss (SIGKILL / kernel panic) as they do from a graceful shutdown. Recovery routines are the primary code path.
4. Zero Unnecessary Allocations
   • In critical data paths, allocations are bottlenecks. Arena allocators, static buffer rings, and amortized capacity keep latency tails (p99.99) predictably flat.

---

📝 Recent Technical Notes & Articles

• Understanding Memory Stalls: Cache Hierarchies and Lock-Free Ring Buffers in High-Throughput Rust
• Lessons from Building an io_uring-Based LSM Engine from Scratch
• Simulating Asymmetric Byzantine and Network Partitions in Raft Clusters with TLA+
• Designing Real-Time Digital Audio Signal Pipelines on Bare-Metal ARM Cortex-M

---

📫 Get in Touch

• Email: julian.vance.dev@proton.me
• PGP Fingerprint: 7E9B 42F1 C8A3 91D0 562B E340 8821 54A0 C672 F18A
• Matrix: @julian:matrix.org
• Code & Specs: https://github.com/ejrniiii8-cloud

---

“Simplicity is prerequisite for reliability.” — Edsger W. Dijkstra