A collection of projects I've worked on, from web applications to system tools and everything in between. Each project represents a challenge solved and skills developed.
A high-throughput encrypted traffic classification system built using CUDA and DPDK, designed to process network flows at over 75 Gbps. Developed for real-time inference and cybersecurity research.
This project focuses on classifying encrypted network traffic in real-time using GPU acceleration. It combines DPDK for high-speed packet ingestion with a custom neural network inference engine written in CUDA. The system targets cybersecurity and traffic engineering applications where line-rate classification is essential.
As part of a distributed systems course at EPFL, me and my group built Dynaster, a Dynamo-inspired key-value store written in Go. The idea was to recreate some of the concepts behind Amazon’s Dynamo — high availability, replication, and fault tolerance — but in a simplified, educational setting.
The project runs on top of Peerster, a decentralized messaging and storage system we had been developing throughout the semester. Dynaster adds a new layer on top of it to manage how data gets distributed, stored, and recovered when things go wrong.
TripTracker is a mobile app we built as a team to make exploring cities on foot more enjoyable and social. The idea was to go beyond maps and directions — to create curated walking itineraries around interesting spots, where users could record their own paths, share them, and discover new ones from the community.
This project was developed as a group effort, and one of the most valuable parts was learning to work as a cohesive engineering team. We split responsibilities across mobile development, backend engineering, and design, but kept a tight feedback loop to integrate everything smoothly.
Every year in Verbier, there’s a limited-time sale offering ski tickets for just 5 CHF.
To avoid missing out, I built a Python bot that automatically checks the website for new ticket availability and sends alerts when tickets appear.
The first version used Selenium and Requests/BeautifulSoup for scraping, connected to Google services to send email notifications.
In the second iteration, I replaced the email system with a Telegram bot, which turned out to be much more convenient and responsive for real-time alerts.
🏆 Winners of the TravelPerk Challenge at HackUPC 2024 — the biggest student hackathon in Europe, held in Barcelona.
This project was done in a team of 3, with Nicolas Font, and Jeremy Kun.
If you want to learn more about the project, check out our Devpost page