AI automation & operations. I take a business task, build the automation around it, and stay responsible for the result.
Portimão, Portugal · EU work authorization · Remote or Algarve · 20–30 h/week LinkedIn · GitHub
Ten years in digital business — performance marketing, affiliate, B2B partnerships, budgets, a team of three developers. The last two years: building and running LLM-based automation.
I am not a model researcher and I do not claim to be one. What I bring is the operator's side of AI: pick the tool that fits, wire it into the business process, ship it, and count what it costs and what it saves.
I am deliberately not looking for model-training or ML-research roles.
A personal AI operations system (2025–2026, ongoing). One laptop, no cloud beyond model calls: an orchestrator that takes a request, a separate executor model that does the work, speech in and out, memory in Markdown, and a runner that sends the same task to several models and records quality, latency and cost per run.
Show its real state rather than a claim. Working: the executor, speech synthesis, live voice conversation, Telegram as a channel, the model-comparison runner. Not working yet: the orchestrator is several contours that have never been joined into one path, notes get lost between chats, the dashboard is down.
Content and publishing pipeline (2026). Scheduled generation, review and publishing for own products through Telegram and a web panel, with logs and idempotent runs.
Before AI (2014–2023). Media buying and affiliate: budgets above $300K, 1,000+ campaigns across Google Ads, Yandex and Meta; affiliate network presence across EU, US, Asia and CIS with 100+ traffic sources; owned PnL analytics (DAU/MAU, retention, LTV, CAC); led a team of three developers.
| Repository | What it is |
|---|---|
| talentflow-agent | Recruiting automation end to end: Djinni parsing, storage, scoring, reply drafts, Telegram digest, 293 tests |
| bookas | Product image synchronisation for a WooCommerce bookstore that uses it |
| ai-lab | Self-hosted Docker stacks for agents, RAG and automation |
Two case studies in this repository: Aspire and Amplify and Texas Handyman.
- Automation with LLMs: prompt and model selection, cost/quality trade-offs, retries, logs, daily budgets, response caching.
- Agent workflows built with AI coding assistants; reading and fixing Python I did not write from scratch.
- API and webhook integration: JSON/YAML pipelines, OAuth, REST, idempotent jobs.
- Voice: speech-to-text and text-to-speech providers, real-time loops, latency and quality measurement.
- No-code / low-code: Flowise, Dify, n8n-style flows, Telegram bots.
- Data: SQLite, CSV pipelines, parsing, simple dashboards; cohort and conversion analysis.
- Business: offer design, pricing, funnel metrics, B2B partner relations, budgets.
Python (with AI assistance) · Docker (running stacks, not orchestrating clusters) · Git and GitHub (day-to-day) · Linux/macOS shell · PostgreSQL · AWS/GCP (basic, exploratory).
- Evidence over adjectives. A claim without a file, a log or a number behind it does not belong in a description of the work.
- A demo that runs beats a feature list that does not.
- Write down what broke. The failure is usually the useful part.
- Written task intake. English in writing is stronger for me than English on a call, and it leaves a trail both sides can check.
Available 20–30 h/week from Portimão. Contract form to be confirmed with an accountant before quoting terms.
Good fit: AI integration, automation of an existing process, rescuing something that stopped working, affiliate and performance marketing analytics.
Not a fit: model training and ML research, pure frontend or pure backend roles, long-term full-time employment.
Reach me on LinkedIn — that is the fastest channel.