Systems & Platform Engineering
Senior Software Engineer/DevOps & Platform/Dublin, IE
Software engineer with 8+ years in telecommunications, going from graduate engineer to senior engineer leading delivery on production systems — Kubernetes, Docker, CI/CD, and cloud-native architecture, built alongside close work with customers on live deployments.
Outside of that, I run a self-hosted infrastructure practice at home: distributed AI orchestration, home automation, an automated content pipeline, and a trading platform — each one built, deployed, and operated solo, end to end. I like problems that require figuring out how something actually works and making it reliable.
| Period | Role | Notes |
|---|---|---|
| May 2021 – Nov 2025 | SR Senior Software Engineer Openet (an Amdocs company) |
|
| Apr 2019 – May 2021 | II Software Engineer II Openet (an Amdocs company) |
|
| Jun 2017 – Mar 2019 | I Software Engineer I → Graduate Openet |
|
| Feb 2016 – Sep 2016 | IN Intern, Services Team Openet |
|
Bloomberg-terminal-inspired full-stack trading platform: FastAPI/PostgreSQL backend, React 19 frontend. JWT auth with Argon2id hashing, race-condition-safe order execution (SELECT ... FOR UPDATE), rate limiting, and an immutable audit trail for security-sensitive actions.
Multi-agent financial advisory pipeline: a Logician, a Devil's Advocate, and an Aggregator agent debate a question in sequence, each running its own system prompt against a locally-hosted model, triggered and delivered via a Telegram bot.
Local-first bilingual (Polish/English) voice & text home assistant. A ReAct reasoning loop running entirely on local hardware via Ollama controls Home Assistant devices and answers questions with cited web search — no cloud dependency for core operation.
Early-stage concept: a fully local, always-on AI presence for the house — broader in scope than HomeAI's conversational control, aimed at ambient awareness of household routines and proactively surfacing what needs attention rather than only responding when asked. Design stage, no repo yet.
End-to-end release pipeline for Husaria Beats: the only manual steps are generating the song and video and a final check. Everything past that — scheduling, cross-posting to TikTok, Instagram, and Facebook, YouTube upload, and cross-linking — runs unattended.
The orchestration and monitoring layer behind the homelab: n8n workflow automation, a Postgres+pgvector store, and a monitoring service that tracks resource usage and can Wake-on-LAN idle compute rather than leaving it running around the clock. See the AI & Homelab Infrastructure case study below for the full picture.
Home Assistant deployment (alarm, gates, lighting, camera-triggered automations) plus Project Nemo, a bespoke aquarium platform built from scratch — reverse-engineered BLE light control, OpenCV + local-vision water test reading, and custom ESPHome firmware, all sharing one real-time source of truth.
A multi-node homelab running an AI-assisted trading platform (FastAPI, TimescaleDB, arq workers), agent orchestration (n8n, pgvector), and a local multi-agent reasoning swarm — all on local-first LLM inference via Ollama, exposed only through Cloudflare Tunnel.
Extending the home automation platform to support Matter, the newer cross-vendor smart home standard, alongside an existing Zigbee device fleet — mapping Zigbee device metadata onto Matter's data model to keep both protocols interoperable during the transition.