Platforms
Multi-tenant products, admin consoles and the APIs behind them. Built to carry many customers on one codebase.
0x.Dev · full-stack engineer
Software that works while I sleep.
I build products end to end: interface, API, payments, servers, and the AI agents that keep them moving.
I’m 0x.Dev, a full-stack engineer who likes owning the whole thing, from the first sketch to the server it runs on.
I’ve built distribution platforms, white-label software, payment systems, real-time apps and a newsroom staffed by AI agents. I run them on infrastructure I set up and harden myself. I move fast, I verify everything, and I write down what I learn so the next fix is quicker.
- alias
- 0x.Dev
- based
- the internet
- role
- engineer, product, ops
- writes
- TypeScript, PHP, Python
- runs
- Linux, nginx, Cloudflare, AWS
- status
- shipping
Selected work.
Six products I designed, built and still run. Client and product names are left out; the work is real.
-
Music distribution platform
Releases go out to every major store, and every cent that comes back gets accounted for.
- Release builder, QC and store delivery feeds
- Royalty imports, splits and payouts in many currencies
- Fraud checks that run on their own
- PHP
- MariaDB
- AWS
- Cloudflare
-
White-label label suite
One codebase that record labels run as their own product, under their own brand and domain.
- Tenant isolation and per-tenant theming
- Custom domains with automatic TLS
- A public API with docs and a changelog
- PHP
- React
- TypeScript
- nginx
-
Checkout and billing
Hosted checkout, invoices and recurring billing, with a kill switch for when something looks wrong.
- Payment links and hosted invoices
- Recurring cycles that bill on time, once
- Add-ons that unlock themselves when paid
- PHP
- Node.js
- Webhooks
- Cron
-
AI newsroom
A multilingual publication where AI desks write, and gates written by a human decide what ships.
- Writing desks with quotas, locks and schedules
- Duplicate and quality gates before publish
- Automatic translation and structured data
- WordPress
- PHP
- LLMs
- Cron
-
Virtual office
A 360° office where the team and AI agents share rooms, desks and group chats.
- Chat and calls with end-to-end encryption
- Presence, movement and room-by-room views
- Agents with their own routines and limits
- React
- TypeScript
- WebRTC
- Node.js
-
Desk tools and pipelines
Small apps that take friction away, and pipelines that publish while nobody is watching.
- A menu bar app, with a mascot, that tracks AI usage limits
- An inbox agent and a clock-in kiosk
- A video pipeline that renders and posts daily
- macOS
- Electron
- Android
- Python
Ship log.
Counted by the servers themselves every fifteen minutes: commits that land on live repos, releases cut, and server changes made by hand.
- ships
- active days
- longest streak
- days in a row
- busiest day
asking the server…
Five rules I don’t bend.
They’re boring on purpose, because boring is what survives contact with production.
-
Read the whole path.
Before I change a line, I follow the request from the click to the database and back. Most bugs live in the hand-offs between the parts.
-
Secure by default, not by phase.
Least privilege, no secrets in code, nothing running as root that doesn’t need to. Security gets built in as I go, so there is never a big clean-up sprint at the end.
-
Ship small, verify live.
Every change goes out in a slice I can roll back in seconds. It counts as done once I’ve watched it work in production, whatever the tests said.
-
Write it down.
Runbooks, decisions, and the trap that caught me last time. The next person to touch this might be me, at 2am, half awake.
-
Automate the second time.
Do it by hand once to understand it. The second time, it becomes a script, a cron job or an agent, with logs and an off switch.
- TypeScript
- PHP
- Node.js
- React
- Next.js
- Python
- MariaDB
- PostgreSQL
- Redis
- nginx
- Docker
- Linux
- Cloudflare
- AWS
- WebRTC
- Electron
Every job, the same four moves.
Big change or one-line fix, it goes through the same loop, and the loop is what keeps it safe.
Map
Read the code, the logs and the data before touching anything. Name every root cause, including the quiet ones.
Build
Small, reversible changes. Boring tech where it matters, sharp tools where they pay for themselves.
Prove
Test in a real browser, against real data, at real screen sizes. Then check it again, live.
Hand over
A short report: what changed, what I verified, what’s still open, and how to roll it back.
What I do.
Mostly the unglamorous middle: the parts that move money, data and messages around without drama.
Payments
Checkout, invoicing, recurring billing, and the reconciling nobody sees.
Infrastructure
Linux, nginx, DNS, TLS, backups. I run what I build.
Automation and agents
Crons, queues and AI agents with guardrails, logs and an off switch.
Real-time
Chat, presence and live dashboards that stay in sync.
Meet Hex.
My rubber duck, upgraded. Hex reads the logs with me, cheers every deploy and dozes off when the cron jobs take over.
or give Hex a tap
Defaults, written down.
If you work with me, this is what you get without having to ask.
- # the defaults, so nobody has to ask
- [working_style]
- read_before_write = true
- change_size = "small, reversible"
- done_means = "verified live"
- secrets_in_code = false
- runs_as_root = "only when it must"
- docs = "every trap, written down"
- reviews = "line by line"
- automate_after = 2 # times by hand
Working together
- Async first. A clear written update beats a long call.
- I keep what I verified apart from what I assume.
- Bad news travels first, because a problem found early is cheap.
- Reports in plain language, so anyone can act on them.
Good software is mostly invisible. It loads fast, fails loudly, recovers quietly and leaves a trail you can follow. I build for the day something breaks, because that day always comes.