Technologies

Skills & technologies

Not a bullet cloud — one honest paragraph per technology: what I did with it, where it ran in production, and how current it is. Grouped by currency, not by topic.

Daily drivers

What I reach for every day right now.

Kimi

AI & Automation

Using Kimi (CLI and API) alongside Cursor for development and workflow assistance — different strengths, same agentic approach.

Claude Code

AI & Automation

Daily driver for development work — agentic coding on real codebases, including large parts of this site. I design the workflow and review the diffs; the agent types fast.

The framework this very site runs on ( Postgres + nginx + gunicorn, custom per-host routing, WYSIWYG editor, SEO tooling ). I hosted Django sites at Sandstone for years, and it's my default choice for anything that needs an admin, a dashboard, and real users.

Linux

Infrastructure

Using Linux since the Slackware days. Ran 24/7 ISP services on it at Interware in the late 90s, trained teams on UNIX at HP Costa Rica, and have run production fleets on Debian/Ubuntu ever since — bare metal and virtualized (Proxmox/QEMU/KVM).

PostgreSQL

Data & Messaging

My default database for years — it backs this site and most of the Sandstone stack. Comfortable with schema design, migrations, tuning the usual suspects, and backups (Restic/rsync).

Python

Languages

My main language since I moved off Java around 2015. Everything on this site is Python/Django; at Sandstone I used it for the special-markets service (FastAPI/RabbitMQ), content-gathering automations (Playwright, Selenium), and the AI rewriting pipelines. Currently my default for LLM workflow glue with n8n, Claude Code, Kimi and Cursor.

Cursor

AI & Automation

Part of the AI-assisted development toolkit alongside Claude Code and Kimi; useful for inline edits and exploration.

WhatsApp (Meta / WABA)

Communications

Integrated both the Meta Cloud API and WABA providers into sites and n8n workflows: inbound message handling, template messages, user verification flows, and content generation from group content.

n8n

AI & Automation

My main workflow orchestrator right now: AI/LLM pipelines, WhatsApp and Telegram integrations, content generation, and verification flows — with Python where n8n's nodes end.

Nginx

Infrastructure

Standard front of every stack I run: reverse proxy, TLS termination, static/media serving, per-host routing. This site sits behind nginx in front of gunicorn.

Communications

Bots and integrations for notifications, contact-form relays (this site pings Telegram on every message), content ingestion from groups, and admin workflows — usually via Python or n8n, or my own sending API.

JavaScript

Languages

Vanilla JS for front-end behaviour (the scroll-fade projects page, drag&drop editors, theme switching on this site are all hand-written). I reach for it when interactivity is needed, without a framework unless one is warranted.

LLM Workflows

AI & Automation

Designing and running LLM pipelines daily: content generation from messy group content, user verification flows, trending-topic discovery feeding rewriting chains. Practical, cost-aware, production-oriented — not demos.

Docker

Infrastructure

Using Docker for local tooling and deployments; comfortable with compose setups, CI runners in containers, and image hygiene.

Content Automation

AI & Automation

Built systems that turn unorganized input (group chats, feeds, provider data) into formatted, AI-written articles: trending-topic discovery feeding LLM rewriters, publishing pipelines, deduplication. Python + n8n + WhatsApp/Telegram sources.

Deep production experience

Technologies I ran or built in production for years.

Asterisk

Communications

Designed an Asterisk/SuiteCRM integration at Sandstone for inbound and outbound messaging with Twilio; earlier wrote Asterisk configs and dialplans at IDSca. Deep PBX experience, mostly historical now that telephony moved to Twilio.

SEO

Legacy & Other

SEO since the affiliate-marketing years (2002–2007): on-page, technical SEO, sitemaps/canonicals/structured data — this site implements the full set, including IndexNow submissions.

Affiliate Tracking

Legacy & Other

Built and ran CPA affiliate systems as an affiliate myself, then implemented inbound/outbound affiliate tracking and performance monitoring at Sandstone — replacing Affiliate Pro with self-written solutions.

FastAPI

Frameworks & Web

Built the 'special markets' (competitions) manager at Sandstone: FastAPI + RabbitMQ with a PHP front-end, mixing provider events with manual entry. My go-to for small, fast service APIs.

MySQL

Data & Messaging

Ran MySQL alongside Postgres for years — LAMP sites, IDSca's stack, hosting at Sandstone. Comfortable with replication, backups, and the usual tuning.

Proxmox

Infrastructure

Virtualized the Sandstone infrastructure on Proxmox: Linux and Windows guests, QEMU/KVM underneath, mail, telephony, workstations, and web hosting all as VMs/containers. Migrated the older KVM setup onto it.

Twilio

Communications

Migrated telephony from on-prem Asterisk to Twilio at Sandstone: voice and messaging flows integrated with the CRM.

BIND / DNS

Legacy & Other

Ran BIND for an ISP at Interware, and automated DNS mass-generation for thousands of white-label domains at IDSca. DNS is one of those things I've debugged at 3 a.m. often enough to respect.

QEMU/KVM

Infrastructure

Ran Windows virtual workstations in QEMU on Debian and managed KVM fleets before consolidating on Proxmox. Comfortable at the libvirt/qemu command-line level.

Postfix / Sendmail / Exim

Legacy & Other

Sendmail and Exim at Interware's ISP in the 90s; Postfix + LDAP mail infrastructure at Sandstone. I no longer volunteer to run mail servers, but I can.

RabbitMQ

Data & Messaging

Ran RabbitMQ as the broker for Sandstone's pipelines for years — queues, exchanges, consumers in Python and PHP. Also ran Mosquitto for MQTT alongside it.

Apache

Infrastructure

Apache since the 90s: 24/7 hosting at Interware, mass-generated vhost configs for hundreds of domains at IDSca, middleware support at HP. These days nginx is my default, but Apache holds no mysteries.

Mosquitto (MQTT)

Data & Messaging

Operated Mosquitto brokers at Sandstone for lightweight pub/sub messaging between services.

Playwright

AI & Automation

Content-gathering automation at Sandstone: Playwright scrapers for JS-heavy sources where plain HTTP wasn't enough.

SQL

Languages

Twenty-five years of SQL across PostgreSQL, MySQL and MS SQL — schema design, query tuning, reporting. It's a constant underneath every stack listed here.

Memcached

Data & Messaging

Used Memcached as the caching layer in IDSca's high-volume sports data infrastructure (Java/ActiveMQ stack).

Selenium

AI & Automation

Browser automation for content gathering at Sandstone before moving the newer scrapers to Playwright.

HA / Load Balancing

Infrastructure

Designed and deployed load-balanced/high-availability systems at IDSca (Linux/middleware layer) for high-volume sports data services.

Hetzner · AWS · Linode

Infrastructure

Managed both cloud and bare-metal fleets across Hetzner, AWS, Linode and Hostinger — provisioning, hardening, backups (Restic, rsync + rsync.net), and cost-driven migrations between providers.

Restic / rsync Backups

Infrastructure

Backup strategy at Sandstone: Restic and scripted rsync to rsync.net for the whole fleet — tested restores, not just green checkmarks.

Working knowledge

Hands-on, but not my main focus — useful when the project calls for it.

HTMX

Frameworks & Web

Completed the HTMX practical guide course and use it where it fits; on this site most interactions are still hand-written vanilla JS.

GraphQL

Frameworks & Web

Built GraphQL APIs with Python (course + practice projects); my production APIs have mostly been REST with Django or FastAPI.

WordPress/CMS

Frameworks & Web

Integrated designs and data into CMS/WordPress front-ends at Sandstone; I treat WP as an integration target rather than a platform I build on by choice.

Kubernetes

Infrastructure

Completed the Docker/Kubernetes zero-to-hero course and run small workloads; most of my production experience is with VMs and plain containers rather than large clusters.

XGBoost

AI & Automation

Personal project: an XGBoost classifier blended with a LogisticRegression model for match predictions. Applied ML rather than research — comfortable with the Python data stack around it.

Historically — happy to dust off

Used heavily in the past; I don't lead with these anymore, but the knowledge is there.