---
title: "About"
canonical: https://solmaz.io/about/
license: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)
---

<div class="about-intro">
<img class="about-intro__photo" src="/assets/images/onur-profile.png" alt="Portrait of Onur Solmaz" loading="lazy" />
<div class="about-intro__text">
<p class="about-intro__lead">Hey, it's Onur.</p>
<p>I'm an engineer. I have a wide variety of interests. </p>
<p>I write about them here.</p>
</div>
</div>

### Current work

I work at [Hugging Face](https://huggingface.co) 🤗, where my current focus is making OpenClaw and other agent harnesses work great with local models.

### OpenClaw

I am a maintainer at [OpenClaw](https://openclaw.ai) 🦞, one of the fastest growing open source projects of all time. I contribute under both [@osolmaz](https://github.com/osolmaz) and [@onutc](https://github.com/onutc).

My work in OpenClaw includes:

- ACP/harness interoperability
- sub-agents
- Claude Code + OpenAI Codex workflow improvements
- a CLI client for ACP
- Discord UX
- MS Teams and other enterprise setups
- security hardening
- browser and terminal use

### Past work

I was a Founding Engineer at [TextCortex](https://textcortex.com), where we helped companies organize their knowledge and build agents. My role involved doing research, designing software architecture, developing new features and leading the engineering team.

While I was trialing work with TextCortex in Fall 2022, I created [the first crappy coding agent](/log/2026/02/13/coding-agent-before-chatgpt/) for them. This was before ChatGPT launched, using a custom Jupyter Kernel and OpenAI's OG Codex code completion model. That crappy CLI coding agent then evolved over time into TextCortex's main chat product.

I was the sole backend developer for 2 years. [Then actual coding agents happened](https://x.com/onusoz/status/1929140559668515195?s=20), and I was able to do a lot more. I later became the lead engineer, overseeing all development.

There were a couple of unique open-source projects we were scaling, one of them is [JSON-DOC](https://github.com/textcortex/JSON-DOC). It's a JSON- and block-based file format for storing documents. Documents are converted into structured data, similar to Notion documents, where each content gets an id, so that AI models can reference and manipulate huge content easier, compared to Markdown.

### Sports

I combine strength training and sports with engineering. See my Instagram channel [Nerd on Bars @nerdonbars](https://www.instagram.com/nerdonbars/).

I am currently at [2 horsepowers](https://www.instagram.com/reel/DTLGhrjiMqf/?igsh=djZzcWhycHowbXl5) of peak muscle-up power.

I am developing a fitness app, [Horse](https://horse.fit), which uses computer vision models and geometric calculations to count reps. We challenge each other with friends, if you are reading this, [I officially challenge you to a 2 minute AMRAP push-up duel](https://horse.fit/@nerdonbars/challenges-you).

<!-- include: github-projects.md -->

### Education

- [Bachelor's in Civil Engineering from METU](https://ce.metu.edu.tr/)
- [Master's in Computational Mechanics from Uni Stuttgart](https://www.commas.uni-stuttgart.de)

### My Accounts

- [Twitter](https://twitter.com/onusoz)
- [GitHub](https://github.com/osolmaz)
- [LinkedIn](https://www.linkedin.com/in/osolmaz/)
- [Hacker News](https://news.ycombinator.com/user?id=hosolmaz)
- [StackExchange](https://stackexchange.com/users/1426841/osolmaz?tab=accounts) (RIP)

### Contact

You can reach out to me at onur@solmaz.io.