Fionn Hughes

2nd year CS @ UCD · Dublin · C++ systems, Python data pipelines
porting PulseDB to Linux

about

Second-year Computer Science student at UCD, based in Dublin. I'm most experienced in C++, I built and shipped PulseDB, a C++ system monitor, on Windows and am now porting it to Linux. I use Python for the open data pipelines below. C and Java from coursework, some TypeScript from open source.

Outside of college I coach tennis at Tennis4Kids, a few evenings a week, running my own groups of under-12s. I've been around the sport for a long time before I ended up coaching.

I like writing code that involves the backend, lower level system code. This page is a log of what I've shipped and what's still being built.

achievements

education

cv (pdf) →

projects

shipped
2026 · v1.0.0

PulseDB

Self-hosted system observability daemon. Shipped v1.0.0 on Windows, now porting to Linux. C++20 backend with its own time-series storage engine, Tauri + React desktop app. Built to replace a Task Manager I don't like.

  • custom .pulse format: LZ4-compressed chunks, delta-encoded timestamps, per-metric WAL with xxHash32 checksums
  • seven native collectors (cpu, ram, disk, network, processes, system, power) on a Boost.Asio scheduler, passed a 24-hour stability run
  • alert engine with threshold, sustained and rate-of-change rules, exposed over a Drogon REST + WebSocket API
  • ships with an NSIS installer; Linux port has storage and four collectors done
c++20 · boost.asio · drogon
sqlite · lz4 · conan
tauri · react
github →
in progress
2026 · wip

Transport Delay Tracker

Logging live GTFS-Realtime data from Dublin's TFI bus network, building toward an ML model that predicts delays before they happen. Working on it with a friend, Viktoriia.

  • polling the NTA GTFS-R feed, writing trip and stop updates to Postgres
  • next: training a delay-prediction model on the logged data
  • self-hosted, running 24/7 on a spare Ubuntu server
python · postgresql
gtfs-realtime
github →
live / auto-updating
2025 · live

Irish open data scraper●

A Python pipeline that pulls a handful of Irish public datasets on a cron, cleans them up, and pushes the output to Kaggle and Hugging Face. Each one has a starter notebook on Kaggle that reruns the day after the data refreshes. Whole thing lives on GitHub Actions, source at FionnHughes/irish-open-datasets.

  • about 2.3M rows in total across the three feeds
  • over 230k combined Reddit views across the three launch posts
  • schedules: PPR monthly, oil bulletin weekly, BER monthly, no manual triggers
currently producing
python · github actions
kaggle · huggingface
open source
2026 · merged

Chess.com piece for Activepieces

A community piece I wrote and got merged into activepieces, an open-source workflow automation platform (think self-hostable Zapier). Wraps Chess.com's public API so workflows can pull player data without needing an API key.

  • three actions: get player profile, get player stats, get daily puzzle
  • Zod validation on the username input, encodeURIComponent on the URL
  • 222 lines added, merged into main as PR #13018
typescript · zod
activepieces
pr #13018 →

out there

now

now

Porting PulseDB to Linux, logging live TFI transit data for a delay-prediction model, and applying for summer 2027 software engineering internships.

  • PulseDB: Windows v1.0.0 shipped. On Linux, storage plus CPU, RAM, disk and network collectors are done; processes, power and the daemon are next
  • Transport Delay Tracker: collecting live GTFS-R data, next step is training the model on it
  • Starting: an Irish Sign Language learning app with camera-based feedback, built with Viktoriia

let's connect

Want a public Irish or EU dataset scraped, cleaned, and republished on a schedule? Get in touch. Small jobs from €10, scales with source complexity. I'll get back to you in a day or two.

email me →

elsewhere