Using OpenCode in real repo work
I made a short about why OpenCode fits my daily workflow: terminal-first, open source, works inside the repo, and cuts down the babysitting.
Watch the video →Hello, I'm
I build software, improve how teams ship, and spend a lot of time figuring out where AI actually helps.
I run Pyango, a software agency in Switzerland. We design, build, deploy, and maintain products for companies that need a strong tech team without hiring one from scratch. Outside client work, I build trading bots, test AI-assisted dev workflows, and simulate biological neural networks in Rust.
The short version: client work at Pyango, a lot of experimenting with developer tools that actually save time, and ongoing trading research.
I made a short about why OpenCode fits my daily workflow: terminal-first, open source, works inside the repo, and cuts down the babysitting.
Watch the video →Most weeks still revolve around client products: design, backend, frontend, deployments, and figuring out how to help small teams move faster without making the process worse.
See Pyango →I keep testing simple quantitative strategies, portfolio approaches, and research workflows. Less hero strategy, more repeatable process.
See the trading work →I started out in a Swiss 4-star hotel kitchen. Good training, brutal hours. But I was already building websites on the side and trying to fix the kitchen's ticket-printer problem with code. Cooking was never going to last.
I taught myself to code and started building MVPs for anyone who'd let me. Over the next decade I shipped dozens of products across web, mobile, and backend. Most of them failed, some of them didn't, and I learned a lot from both.
I launched Pyango as a software agency and AI consultancy. Subscription-based, pausable, built for companies that need a full tech team without the overhead. Grew to 8 people spread across Switzerland, Germany, Iran, and Tunisia.
I run Pyango while going deep on AI-augmented development. We use agentic workflows across the team, build AI-powered tools for clients, and I still find time for trading bots and neural network research in Rust.
This is where most of my energy goes right now. I build practical AI workflows for my team and our clients: code review helpers, research agents, faster testing loops, and less copy-paste work. I also help teams figure out where these tools save real time and where they just create noise.
Read more →My software agency and AI consultancy. We work with businesses as their outsourced tech team. Design in Figma, build in Django and Vue, deploy, maintain. Subscription model, pausable monthly. 8-person international team across 4 countries. We handle everything so our clients can focus on their business.
Read more →A JetBrains IDE plugin I built at Pyango. It lets you manage GitLab CI/CD environment variables directly from your IDE instead of clicking through the GitLab UI. Works with PyCharm, IntelliJ, WebStorm, and every other JetBrains IDE. It is at 594 downloads right now.
Read more →Quantitative crypto strategies built with Python and the Jesse framework. I run walk-forward validation with VectorBT Pro. The best strategy hit +118.7% on BTC with a Sharpe of 2.17. The idea is lots of simple strategies running at once instead of chasing one perfect system.
Read more →Start with the customer's pain. Validate before you build. Every feature you skip is time saved for the one that actually matters.
Trading bots, agentic workflows, data scrapers. I want to build the machine, not be the machine. If I'm doing the same task twice, I'd rather spend 10x the time automating it.
Multiple simple trading strategies beat one perfect one. Multiple small businesses beat one fragile bet. I'd rather have a portfolio of okay bets than put everything on one moonshot.
From the Cult of Done: "Pretending you know what you're doing is almost the same as knowing what you're doing." I ship things before they're ready, and then I fix them. Perfection is a trap.
The actual repo behind a lot of what I build, test, write about, and automate. Projects, experiments, trading work, notes, and random side quests.
StarterA starting point I give away for free. Useful if you want a practical base instead of staring at an empty repo and pretending that is part of the fun.
VideoA short look at the setup I keep coming back to: repo-first, terminal-first, flexible providers, and less context switching.
WritingLonger posts on product work, software delivery, AI workflows, and what we learn while building things for clients.
I make videos about product work, dev workflows, and what it's actually like to run a small tech agency.
A straightforward breakdown of how we structure CTO-as-a-service work at Pyango and where it makes sense for a company.
A short on the setup I use when I want less babysitting, less tab switching, and more actual work getting done inside the repo.
Things I'm exploring or building right now. Some will ship and some will die on the vine. That's fine.
Building internal tools that cut busywork for our team. Better code review loops, better handoffs, less context switching, and fewer boring tasks eating up the day.
The idea: find fragmented public data, scrape it all into one place, and monetize the traffic through SEO and ads. WasteCal is the first one. I'm looking at more verticals to apply this to.
A real estate insights tool. The idea is to help people make better property decisions by pulling together data that's normally scattered across ten different sites.
A platform that scrapes public events from across the internet and puts them on a map. Like Google Maps but for "what's happening near me this weekend."
If you want to talk about software, better dev workflows, trading bots, neuroscience, or a project you are building, send me a message.