Neuromorphic KV store in Zig

Source: Dev.to

Introduction

Hello! In 2026 I decided to start a new project in Zig: a neuromorphic DB engine called Walbash DB. The name is a play on Wabash, the neighborhood where I grew up in Kansas City, and on the concepts WAL (Write‑Ahead Log) and Bash (Bourne Again Shell). It’s a silly name, but it stuck.

I’m building a game and want an excuse to explore neuromorphic computing. The database will store synaptic weights for a neuromorphic engine I’m working on.

I’m not a “smart guy” in the sense that I prefer low cognitive load. In my engineering career I’ve found that languages that front‑load many features (e.g., Scala, Rust) make me less productive. I work with Go professionally, but I wanted something lower‑level that gives me maximum control while remaining cognitively straightforward. Zig fit the bill. Odin is also low‑level, but its syntax feels too similar to Go for me, making it hard to break the habit of writing Go‑style code. Zig introduces a different muscle memory, which I’ve always wanted to develop.

Learning the supporting theory for a database has been challenging. I’ve read books on databases and often found myself copy‑pasting code without fully understanding the “why.” AI has become a useful research partner for me, allowing me to learn at my own pace and in my own style. However, I’m aware that the performance goals I’m targeting are too low‑level and niche to trust AI output completely, and I don’t want to lose the joy of learning and implementing things myself.

This project is purely a learning exercise. I don’t expect competitive features, and it’s intended for personal use only. I’m blogging about it to share my journey and keep myself on track.

Progress is fast at the start as I set up the DB, but I expect it to slow down once I get into memtables and testing—building a DB from scratch is hard. I’ll be splitting my time between the game, the neuromorphic engine, and this DB, all on a part‑time basis.