Projects

logilinux 1st Place Overall Logitech Challenge Special Mention JetBrains Challenge 4th Place
Linux HID++ Implementation & Support for the Logitech Master Series We decided to fill a major pain-point of the Logitech Master Series, and Logi Options+: The lack of Linux support. Being low-level engineers, and also linux users, we found that this ecosystem had little to no linux support, and any open source solutions were quite-frankly lacking. Logi Options+ provides a lot of customization and features that no available solutions on linux do, and Logitech's proprietary software does not currently support linux. Hence, we decided to reverse engineer the protocol on some of the most obscure devices in this series: The MX Creative Console pack. This version is mainly centered around adding as much support for these devices on linux as possible, while also providing a development kit for other linux users to create innovative solutions and uses for these devices without having to go through the endeavors that we did. We have implemented pretty much all of what Logi Options+ lets you do with these devices on linux, and hope Logitech appreciates it :D
Arjun Juneja, Rudrrayan Manna
Link
KnowTube 2nd Place Overall
KnowTube turns YouTube videos into interactive lessons with auto‑generated quizzes and flashcards KnowTube is a learning companion that transforms YouTube videos into structured, active study materials. Instead of passively watching, learners can upload or link a video and instantly get a clean transcript, comprehension quizzes, and spaced‑repetition‑friendly flashcards generated from the key ideas in the content. Behind the scenes, KnowTube analyzes the video’s transcript to detect concepts, definitions, and relationships, then builds question sets and flashcards that encourage recall rather than recognition. Quiz routes let users test understanding right after watching, while flashcard routes help them review and retain what they’ve learned over time. Designed for students, educators, and self‑learners, KnowTube makes it easy to turn everyday videos into reusable micro‑courses without manual note‑taking or question writing.
Dominik Glandorf, Fares Fawzi, Seyed Parsa Neshaei, Abhinand Shibu
Link
showcAIse 3rd Place Overall Huawei Challenge 2nd Place
AI-powered presentation analysis tool that provides instant feedback on your speaking performance. AI-powered presentation analysis tool that provides instant feedback on your speaking performance. Upload a video and get detailed insights on delivery, pacing, confidence, and areas to improve. Will also provide a generated audio giving an example of how you should ideally present.
Fatumah Binta Sakho Doukouré, Oscar de Francesca, Ismail Nour, Kristyn Dede Korboe
Link
Birdy Organizers' prize
Posture tracker and hand-gesture controller that runs quietly in the background Birdy uses your webcam to monitor posture and detect hand gestures in real time. It runs in the background, alerts you when you slouch, and lets you control actions using simple gestures. Designed to improve comfort, reduce strain, and enhance productivity while you work.
Sylvain Pichot
Link
PawPilot Huawei Challenge 1st Place
Provides real-time obstacle detection and environmental awareness for visually impaired individuals Mobile app which provides real-time obstacle detection and environmental awareness to enhance independence for visually impaired individuals.
Lorenzo Marogna, Marcelina Gancewska, Wissam Pheng
Link
LLM Inference Optimizer Huawei Challenge 3rd Place
Improving inference of LLM Improving inference of LLM by doing better KV caching
Reinatt Hansel Wijaya, Sathvik Bhagavan
Link
MX Sense Logitech Challenge Winner
Helping visually impaired people navigate their computers using haptic feedback Using haptic feedback in different ways to help visually impaired people navigate and find buttons, input fields, etc. There is a large market segment that we can sell mice to and thus capture new users.
Arthur Brunborg
Link
3DX Logitech Challenge 2nd Place
Exploring 3D Images with seamless integration using Logitech's equipment Exploring 3D Images with seamless integration using Logitech's equipment
Matteo Caviglia
Link
JARVIS Logitech Challenge 3rd Place
A hands-free AI assistant that sees what you see, hears what you say, and responds naturally through JARVIS is an intelligent multimodal assistant that creates seamless human-AI interaction through voice and vision. Speak, and the system automatically captures your webcam feed, transcribes your speech, and sends both to GPT for contextual understanding. The AI responds with natural spoken audio, creating a fluid conversation that understands both what you're saying and what you're looking at. Built entirely with OpenAI's latest APIs, JARVIS demonstrates the future of accessible, hands-free AI assistance—perfect for scenarios where typing isn't possible, from accessibility applications to hands-busy work environments.
Aaron Dinesh, Carlos Hurtado, Guillaume Krafft, Krish Chawla
Link
Gestures plugin for Jetbrains IDE JetBrains Challenge Winner
Navigating through the code in the IDE using gestures. Navigating through the code in the IDE using gestures. Uses computer vision to recognize what gesture is user showing to the camera and executing commands in the IDE. Tracking of the eye movement and mapping where on the screen is user looking. On hand gesture the code that is observed is being sent to the Jetbrains AI requesting the explanation of the code.
Ognjen Gligorić, Nemanja Stjepanovic, Nenad Berić, Stefan Durica
Link
Exceptional Assistant Jetbrains Challenge 2nd Place
Intelligent Coding Assistant – Focused on Error Detection and Resolution Intelligent Coding Assistant – Focused on Error Detection and Resolution An AI-powered development tool designed to help you write cleaner, more reliable code. Instantly detects syntax errors, runtime exceptions, and potential bugs, providing actionable explanations, suggested fixes, and code snippets to resolve issues faster. Perfect for improving productivity and reducing debugging time.
Alexandre Benzonana, Daniel Rainho, Stefano Cirieco, Tiago Gama
Link
BrainsGraph JetBrains Challenge 3rd Place
We map codebase investigations onto an interactive graph for personalized learning. Have you ever tried to onboard to a really large codebase? It's incredibly hard. What about using AI agents? They'll generate massive summaries of your entire codebase. But that's rarely helpful—it's almost equivalent to reading the entire codebase yourself. We want something more visual and interactive. We're building a tool that visualizes codebases as interactive graphs. You ask dozens of questions—"Where is authentication handled?" "What depends on this module?" "How does data flow through the system?"—and each time, the tool highlights the specific regions of the graph involved, along with clear text explanations. By the end, you have a color-coded graph that reflects all your investigations: a personalized map of exactly what you explored and learned.
Adam Chahed Ouazzani, Rassene M'sadaa, Mohamed Mokhtar Sellami, Youssef Boughizane
Link
NodePass SBB Challenge Winner