Not a list of buzzwords. These are systems we've built and run in production. Each one was born from a real need, tested under live conditions, and refined through iteration.
Container orchestration with Docker, reverse proxy with Traefik v3. Every service runs in its own container with automatic TLS certificates via Let's Encrypt. No manual SSL renewal, no port conflicts, no 3am phone calls.
Static site hosting with nginx:alpine. Custom configs, gzip compression, cache headers, and try_files for clean URLs. The website you're reading runs on this stack.
Linux server administration: SSH hardening, firewall rules, backup strategies, performance monitoring. We manage the full stack from DNS to database, without managed hosting platforms.
Visual workflow automation connecting APIs, databases, and services. From fetching broker data to triggering agent pipelines. n8n is the glue that holds our systems together without writing custom integrations every time.
An automated pipeline that watches a Nextcloud folder, names incoming documents based on content, routes them to the right recipient, and files them in the correct archive. Drop a PDF in the folder, it gets sorted, named, and forwarded. No manual filing.
A private variant: documents arrive via scan-to-mail, get read by GPT-4o (OCR + content extraction), and are automatically named and filed on the cloud. From paper to organized archive without touching a keyboard.
Time-based automation: heartbeat monitoring, periodic data fetches, scheduled reports, and agent wake-up calls. Every recurring task runs on its own schedule, with Telegram alerts on failure.
Agent orchestration with persistent memory, sub-agent delegation, and tool integration. THEA runs on Hermes, she coordinates other agents, manages infrastructure, and handles content. Not a chatbot. A persistent system.
Multi-model LLM access via OpenRouter and Ollama Cloud. Prompt engineering, structured output (JSON), and model selection based on task. Trading decisions use one model, review and analysis use another. Each picked for its strengths.
Multi-agent systems where agents delegate, communicate, and persist knowledge. ZETA trades, THEA manages, CLEO handles content pipelines. Each agent has its own role, memory, and runtime, but they share infrastructure.
Airtable alternative, self-hosted. CRUD APIs, relational fields, and a clean UI. ZETA's decision log, position tracking, and win-rate calculations all live in Teable. No external database service.
File sync, sharing, and collaboration. Self-hosted. The backbone of our document management pipelines. Dropzone folders feed into automated workflows that name, route, and archive files.
Custom Python scripts for data processing, API clients, and automation glue. From normalizing broker API data to computing rolling win-rates. Python is the language between the APIs and the database.
Telegram bots as the human interface to our agents. THEA communicates via Telegram: alerts, commands, daily briefings, and interactive conversations. One app, full control.
Automated content: social media scheduling, YouTube uploads with metadata, and portfolio site updates. CLEO handles the full pipeline from Nextcloud dropzone to published content. No manual uploads.
Every skill here was built because we needed it, not because it was trendy. The toolbox grows when the problem demands it.