achu

i build , , and .

LOCAL RUNTIME
MODEL IO
CORE TEMP 37C
Achu runtime core visual state: default.
LOCAL RUNTIME / INFERENCE CORE
CONTEXT WINDOW / AGENT TRACE / MEMORY GRAPH
active layeridle / composed

Focus areas

I like making accessible systems that excite and immerse you.

Local AI infrastructure, developer tools, creative technology, and web experiences that feel alive.

01

local-first AI

Interfaces and gateways that make local models feel fast, inspectable, and useful in real workflows.

02

developer systems

Agent loops, protocol layers, CLIs, and product surfaces built for people who ship software.

03

creative tooling

Memory, workflow, and context systems for making creative output compound instead of scatter.

04

experiential web/worlds

Cinematic web surfaces, music systems, live events, and worlds that feel inhabited.

Projects

Things I've built

View all work
MODEL GATEWAY LOCAL RUNTIME

Relay

OpenAI/Anthropic-compatible shim for local llama.cpp inference.

Turns fragmented local inference APIs into a stable agent target.

TypeScriptSSEllama.cpp
open module
CODING AGENT AGENT TRACE

Synax

Terminal-native agent loop for quantized local and cloud models.

Makes the agent loop inspectable while code is being changed.

TUIProvidersTools
open module
MEMORY OS MEMORY GRAPH

wytOS

Structured creative memory with SQLite FTS5 search and local LLM assistance.

Keeps creative context available instead of scattered across tools.

RetrievalAutomationContext
open module
AUDIO WORLD WORLD IO

watchyourtemper

Electronic music, events, merch, and worldbuilding.

Connects releases, events, content, and commerce into one world.

WorkersMusicCommerce
open module

Background

Fullstack/backend engineer based in Toronto.

Fullstack EngineerBackend EngineerAI ToolsLocal LLM SystemsEdge / Serverless

I build across model gateways, coding agents, creative operating systems, edge-deployed applications, and self-hosted infrastructure. The throughline is local-first architecture, fast feedback loops, and interfaces that reveal how the system behaves.

Resume

Writing

Notes and devlogs.

Notes on local inference, agent observability, protocol design, creative tooling, self-hosted infrastructure, and the build process behind this portfolio system.

Notes