Full-stack developer. System tinkerer. Local AI enthusiast.

I like building things end-to-end — from the UI and API to Docker, deployment, infrastructure, and lately, the AI running beside the code.

Most of my work lives somewhere between web engineering, DevOps, and local AI.

code → break it → understand it → rebuild it better

My current rabbit hole.

OwA is an open-source, local-first coding agent built around Ollama.

The goal is simple:

Your model runs on your machine.
Your code stays on your machine.
Your agent should actually understand your workspace.

It combines repository context, semantic search, tool execution, Git awareness, and local LLMs to make smaller models genuinely useful for software development.

→ github.com/ZakaCoding/ollama-workspace-agent

pip install ollama-workspace-agent
owa

No cloud required. No telemetry. No external AI API keys.

Just your machine, your model, and your code.

Backend & Web

PHP · Laravel · CodeIgniter · Python · JavaScript · TypeScript

Infrastructure

Docker · Linux · Nginx · Redis · Git · CI/CD · AWS

Data

MySQL · PostgreSQL

Local AI

Ollama · LLM Agents · RAG · Embeddings · Tool Calling

I'm especially interested in making small local models more capable instead of assuming every problem needs a massive cloud model.

My GitHub is basically a timeline of things I wanted to understand badly enough to build.

I've worked on things ranging from:

• full-stack business applications
• logistics and operational systems
• e-commerce
• real-time chat
• research tools
• CI/CD and infrastructure
• local AI infrastructure
• coding agents and retrieval systems

One older project I'm still fond of is Open CMAP, an open-source reflective concept-mapping tool originally built from my college research.

→ github.com/ZakaCoding/open-cmap

Some experiments become products.

Some become abandoned repositories.

Some teach me enough to start the next one.

That's part of the fun.

I tend to prefer:

local > cloud, when local makes sense
simple > clever
working software > beautiful architecture diagrams
open source > black boxes
understanding the system > just making the error disappear

I don't stick to one layer of the stack.

If something breaks between the browser, API, database, container, reverse proxy, network, or model...

that's probably where I'll end up digging.

Right now I'm spending a lot of time around:

local LLMs · coding agents · context retrieval · RAG · small-model tool use · developer infrastructure

I'm particularly interested in the question:

How capable can a coding agent become when everything runs locally?

OwA is my attempt to find out.

I build because I'm curious.

If something I build becomes useful to somebody else, even better.

Feel free to explore the repositories, open an issue, start a discussion, or break something and tell me about it.

Zaka Noor
Full-stack developer
Builder of OwA 🐒
Usually somewhere between code, containers, and a local LLM.

🌐 zakacoding.github.io