“I worked with Robin on a game at the Global Game Jam - Belgium. I was impressed with his work on the game, cranking out the basic gameplay in the short time that was given to us. For a young programmer he performed admirably and well beyond my expectations, delivering great code to get the game working.”
About
Web Protocol and Performance expert at Akamai.
PhD in Computer Science.
Google…
Activity
-
De vierde lichting studenten komt eraan! Ook frontender worden? Of ken je iemand die hiervoor gemaakt is? Op zaterdag 13 december houden we weer een…
De vierde lichting studenten komt eraan! Ook frontender worden? Of ken je iemand die hiervoor gemaakt is? Op zaterdag 13 december houden we weer een…
Liked by Robin Marx
-
Talking to Henri Helvetica later today! About web performance or something!
Talking to Henri Helvetica later today! About web performance or something!
Liked by Robin Marx
-
Small improvement to our request waterfall: view the request count and page weight for your selected requests. For example, you can quickly see how…
Small improvement to our request waterfall: view the request count and page weight for your selected requests. For example, you can quickly see how…
Liked by Robin Marx
Experience
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Education
-
UHasselt
-
Distinction with special congratulations from the Jury
"Debugging Modern Web Protocols"
A study of HTTP/2, HTTP/3 and QUIC, their use in practice and their impact on web page loading performance.
Main contributions:
- Analysis of best practices in resource prioritization and multiplexing
- Creation of analysis frameworks and tools (qlog and qvis) that are used internally by Facebook, Cloudflare, Apple, Amazon, ...
- Debugging and bug root cause analysis of dozens…Distinction with special congratulations from the Jury
"Debugging Modern Web Protocols"
A study of HTTP/2, HTTP/3 and QUIC, their use in practice and their impact on web page loading performance.
Main contributions:
- Analysis of best practices in resource prioritization and multiplexing
- Creation of analysis frameworks and tools (qlog and qvis) that are used internally by Facebook, Cloudflare, Apple, Amazon, ...
- Debugging and bug root cause analysis of dozens of protocol implementations -
-
-
-
-
-
-
-
-
-
-
-
Volunteer Experience
-
Trainer
Invirtus Historical European Martial Arts
- Present 6 years 11 months
Education
I teach classes 2 to 3 times a week in Historical European Martial Arts with a focus on longsword
Publications
-
Real-time Hand-Painted Graphics for Mobile Games
AMDO 2014 / LNCS
This paper received the "AMDO 2014 Most Suitable for Commercial Application" Award.
In this paper we set out to find a digital painting technique which allows the recreation of scenes from childrens books or graphic novels for usage in mobile games. The idea is to give the impression that scenes are being hand-painted at the moment the player is moving through them.
To this end, we propose a technique for digital painting, mesh-based strokes, that combines the strengths of…This paper received the "AMDO 2014 Most Suitable for Commercial Application" Award.
In this paper we set out to find a digital painting technique which allows the recreation of scenes from childrens books or graphic novels for usage in mobile games. The idea is to give the impression that scenes are being hand-painted at the moment the player is moving through them.
To this end, we propose a technique for digital painting, mesh-based strokes, that combines the strengths of existing painting techniques while keeping the limitations of mobile devices in mind. In this new technique, instead of representing strokes as pixels in a raster, strokes are created as flat meshes following the contour of a brush stroke.Other authors -
-
An Evaluation of the Impact of Game Genre on User Experience in Cloud Gaming
IEEE Games Innovation conference (IGIC) 2013
Cloud gaming, in which the processing power of a
datacenter-based infrastructure is utilized versus local resources,
is a popular topic in research. This technology is successfully
applied by vendors to enable low-end hardware to deliver a
similar gameplay experience to state of the art consoles. Many
works in literature have focused on the quantitative aspects of the
technology (i.e. delay measurements, visual quality determination
etc), but the qualitative factors have…Cloud gaming, in which the processing power of a
datacenter-based infrastructure is utilized versus local resources,
is a popular topic in research. This technology is successfully
applied by vendors to enable low-end hardware to deliver a
similar gameplay experience to state of the art consoles. Many
works in literature have focused on the quantitative aspects of the
technology (i.e. delay measurements, visual quality determination
etc), but the qualitative factors have not received a similar
systematic treatment. Games are typically classified in terms
of their gameplay into a distinct category or genre, including
action, puzzle, strategy and racing games. In this work, a
qualitative comparison of these genres is presented based on
a common testing methodology which combines both objective
(based on physiological measurements) and subjective (based on
user evaluation) approaches. While in normal networked games,
only multiplayer experiences are subject to the detrimental effect
of delay, the nature of cloud gaming may result in an impact
on singleplayer experiences as well. Results from this analysis
hint at the fact that there is a similarity in delay-sensitiveness
over the different genres in both singleplayer cloud gaming setups
and traditional networked multiplayer games. More in particular,
results show that action-oriented games are more sensitive to
network delay in both setups when compared to other genresOther authors -
-
Overlay Network Based Optimization of Data Flows in Large Scale Client-Server-based Game Architectures for Deployment on Cloud Platforms
Proceedings Terena TNC2012 - Networking to Services
In this overview paper, we present results that were obtained by combining a specific architectural design for Networked Virtual Environments (NVEs) with a generic optimization infrastructure for content streaming. Although the two are –at first sight – targeting unrelated application areas, it turns out that the application knowledge available in the NVE architecture can effectively be used to instruct the adaptation infrastructure to change the way in which data is delivered to the end-user…
In this overview paper, we present results that were obtained by combining a specific architectural design for Networked Virtual Environments (NVEs) with a generic optimization infrastructure for content streaming. Although the two are –at first sight – targeting unrelated application areas, it turns out that the application knowledge available in the NVE architecture can effectively be used to instruct the adaptation infrastructure to change the way in which data is delivered to the end-user. The end- effect of these optimizations is a fine-grained ability for the combined architecture to deliver the required NVE content to the end-user while optimizing bandwidth usage. Such optimizations are very useful in mobile setups, as the available bandwidth fluctuates significantly over time. The resulting architecture can be deployed in a very dynamic way, including on cloud platforms. An experiment is presented that shows the viability of the ideas discussed in the paper.
Other authors -
Projects
-
Runes
- Present
Runes is a combat game in which you play a druid fighting for political power. The player uses the Leap Motion to draw magical runes on the screen and cast them on the enemy player or an NPC.
The player has to learn the runes, their powers and weaknesses. To win, you need the right strategy and perfect timing; it's like playing a cool, expanded, fast-paced version of rock-paper-scissors. Leap Motion is, for us, the perfect technology to make this game as exciting as possible and make…Runes is a combat game in which you play a druid fighting for political power. The player uses the Leap Motion to draw magical runes on the screen and cast them on the enemy player or an NPC.
The player has to learn the runes, their powers and weaknesses. To win, you need the right strategy and perfect timing; it's like playing a cool, expanded, fast-paced version of rock-paper-scissors. Leap Motion is, for us, the perfect technology to make this game as exciting as possible and make the player really feel like he is actually casting magic, instead of just pressing a button to do so.Other creators -
-
Troades Digital Signage
-
For this project, I developed a custom front-end "media player" for Digital Signage content (served by a CMS) using Unity3D. Content is downloaded and updated on-the-fly, without the user being aware, providing a seamless signage experience. In the future, the system will be expanded with Augmented Reality and social networking integration.
The main challenge was to create a robust build for all platforms (Windows, OS X, iOS and Android), especially given the requirement of video…For this project, I developed a custom front-end "media player" for Digital Signage content (served by a CMS) using Unity3D. Content is downloaded and updated on-the-fly, without the user being aware, providing a seamless signage experience. In the future, the system will be expanded with Augmented Reality and social networking integration.
The main challenge was to create a robust build for all platforms (Windows, OS X, iOS and Android), especially given the requirement of video playback. Since the application has to run for hours at a time without restarting, preventing and fixing memory leaks was an important part of the project. -
Kika And Bob
-
In this project, LuGus Studios created 26 unique minigames, to coincide with 26 episodes of the animated series Kika and Bob. The games are based on 6 different game types (endless runner, pacman/bomberman, diner dash, frogger, guitar hero and whack-a-mole), but each game has unique graphics (taken from the episodes) and specific changes to gameplay.
The games were first released individually on various tv-channel websites (including Ketnet.be) and later bundled as a single iOS app, adding…In this project, LuGus Studios created 26 unique minigames, to coincide with 26 episodes of the animated series Kika and Bob. The games are based on 6 different game types (endless runner, pacman/bomberman, diner dash, frogger, guitar hero and whack-a-mole), but each game has unique graphics (taken from the episodes) and specific changes to gameplay.
The games were first released individually on various tv-channel websites (including Ketnet.be) and later bundled as a single iOS app, adding highscores, custom avatar, and a 27th tower-defense game.
Robin was responsible for the project's technical setup and principal architecture, overall project management, resource pipeline, gameplay prototyping and design, implementation of gameplay and algorithms.
Other creators -
-
JIMMobile Spank The Pony
-
Spank the Pony is a carnival-like installation where players take place on mechanical horses that are hooked up to a digital game. By "riding" the horse and spanking its bottom, they try to be the first to reach the finish.
I developed the networking interface (UDP-based) between the PC game and the horses (PLC-based controller), implemented the gameplay and provided the technical hardware setup and support.Other creators -
-
Project Baxter
-
A state of the art prototype for an educational and therapeutic game for a multinational client in the heallth sector. The game was designed to be as accessible (multi-platform) and user-friendly (young children) as possible.
Other creators -
Languages
-
English
Native or bilingual proficiency
-
Dutch
Native or bilingual proficiency
-
French
Limited working proficiency
-
German
Elementary proficiency
-
Latin
Limited working proficiency
Recommendations received
3 people have recommended Robin
Join now to viewMore activity by Robin
-
This was something that might not happen again: presenting in Lederhosen! I was invited by Shopware to speak at their Shoptoberfest Benelux edition.…
This was something that might not happen again: presenting in Lederhosen! I was invited by Shopware to speak at their Shoptoberfest Benelux edition.…
Liked by Robin Marx
-
The DebugBear request waterfall back when it was still built on top of Highcharts 😂
The DebugBear request waterfall back when it was still built on top of Highcharts 😂
Liked by Robin Marx
-
This is one of those hidden secrets where the “easy” head vs body abstractions of the web break because browsers don’t behave in a simple way and…
This is one of those hidden secrets where the “easy” head vs body abstractions of the web break because browsers don’t behave in a simple way and…
Liked by Robin Marx
-
The reception for the game has just been beyond amazing!
The reception for the game has just been beyond amazing!
Liked by Robin Marx
-
Helena launched SpeedCurve 12 years, 1 month and 5 days ago, and today we sold it to US company Embrace. It’s been a heck of a ride that exceeded…
Helena launched SpeedCurve 12 years, 1 month and 5 days ago, and today we sold it to US company Embrace. It’s been a heck of a ride that exceeded…
Liked by Robin Marx
-
Recorded a video explaining how to use WebPageTest and interpret the request waterfall!
Recorded a video explaining how to use WebPageTest and interpret the request waterfall!
Liked by Robin Marx
Other similar profiles
Explore more posts
-
Addy Osmani
New in Chrome DevTools: Annotate performance findings with Gemini! We just shipped a brand new feature in DevTools. You can now ask Gemini to explain performance traces. Specifically you can generate annotations about events in the performance trace. This is a total game-changer. Double-click an event in the Main track, then click Generate label next to the input field. Gemini can suggest a label based on the stack trace and context. Try it out and let us know what you think! #ai #programming #softwareengineering
54 Comments -
Gunnar Morling
"Why do #Debezium data change events have a 'before' part?" Have you ever wondered why there is not only the new state of a row represented in Debezium events, but--for updates and deletes--also the old state? Indeed this is not required for all CDC use cases; for instance, when building a realtime ETL pipeline to push data into an OLAP store or a search index, applying the current state, using upsert semantics, is all that's needed. In scenarios like this, having the 'before' part just increases the message size, also it may increase the size of the WAL in the source database (case in point, Postgres admins may be reluctant to enable replica identity FULL, which is needed for capturing the old row image). But there's one important case though where having the 'before' part is vital: stateful stream processing with tools like Flink SQL, specifically operators like aggregations. As an example, consider a streaming query which uses the SUM() function in a GROUP BY aggregation. When updating one of the values feeding into that aggregation, a retraction of the previous value is required in order to perform the computation. This is where the 'before' part of the Debezium events comes in: it provides the previous value of each field of an updated row, thus enabling the incremental computation of functions like SUM(), solely based on incoming data change events. Internally, Flink splits up the Debezium events into two: a retract message (denoted as "-U") and an add message ("+U"). When using Flink SQL on an upsert changelog stream (i.e. one that is lacking retract events), the query planner can optionally add a ChangelogNormalize operator for materializing retract events. This is costly though in terms of state size: in order to do so, that operator needs to keep the previous state for each key. Having the 'before' part in the Debezium events avoids that issue. Finally, when using the Flink CDC connector for Postgres (which is based on Debezium) and you don't actually need the old row image in your pipeline, you can set 'changelog-mode' to 'upsert', in which case you can use the default replica identity for your tables. I hope this helps to clarify the value of the 'before' part of Debezium's change events, and when to use it (or not). Like so often the case with stream processing, things which seem trivial on the surface can be a surprisingly nuanced matter actually.
14 Comments -
Umar Lone
🚀 New Blog Post! 🚀 I’ve just published a deep dive into a C++20 feature that balances safety and performance: make_unique_for_overwrite and its sibling make_shared_for_overwrite. This article untangles the real-world differences from classic memory allocation patterns in C++, explains when and why to use these functions, and highlights how to avoid the pitfalls of uninitialized memory—with insights across compilers. If you strive to write safer, faster C++ or want to strengthen your grasp of C++20’s evolving foundation, this read is for you. https://lnkd.in/dA7DxD7u 🔗 Check it out, and let me know your thoughts and experiences—especially if you’re applying these patterns in production code! #cplusplus #cpp20 #programming #moderncpp #softwareengineering #memorymanagement
1 Comment -
Davy Demeyer
Ok now I feel bad about bashing on Industrie 4.0 a couple of posts back ... 🙈 A German friend just pointed out there's an MIT licensed AAS Repository available for anyone to play with. (AAS = Asset Administration Shell. Basically the thing that will move us from siloed systems to an open ecosystem. Single data entry. Only Once. Streamlined engineering. Maybe the github for industrial engineering ...?) And they have SDK's in Python, TypeScript, Java, .Net and Rust! 👀 Web UI built with Vue.js 💚 And cool demo's for converting old pdf data-sheets to AAS using LLMs 🥷 And all running in Docker 🚢 . (Yes this is unfortunately still special in industry ...) github: https://lnkd.in/gaFanfPa Like and share to create much more awareness for this cool work! 🫵
56 Comments -
Jan Ozer
Is InterDigital’s Deep Render Deal the First Step Toward an AI-Native Codec Standard? InterDigital’s acquisition of Deep Render marks one of the first times a traditional codec IP powerhouse has bought into AI-native compression. Deep Render was founded on a clean IP picture that doesn’t trace its roots back to 30-year-old block-based video architectures. At some point, that architecture will be abandoned. Whether Deep Render’s technology is what finally replaces it is too early to tell, but InterDigital is making an interesting bet. Depending on how this plays out, they could: • inject some of Deep Render’s AI work into H.267 • push to make H.267 itself an AI-native codec • or build a parallel, proprietary AI-based track altogether If the next generation of codecs turns out to be AI-first, InterDigital will already have both the patents and the people. If not, they’ve still absorbed one of the few teams that proved AI-based compression could work beyond slides and demos. For anyone looking to catch up on Deep Render’s technology, here’s my report: https://lnkd.in/e7qXGi38 You can see my demo of the Deep Render FFmpeg-based workflow in this video: https://lnkd.in/ebGFyjmx The company's demo here: https://lnkd.in/eNMrA3fj And an interview with CEO Chris Besenbruch here: https://lnkd.in/ebxwfWbV The purchase price wasn’t disclosed, but the signal is clear: Deep Render achieved enough technical and intellectual property milestones to make a company like InterDigital take notice. That’s an achievement in itself. Congrats.
1 Comment -
John Paulett
My Pair Programming with AI workflow at HOPPR turning vibes into real code (with a few tricks 🧠) - Draft a plan file. I usually write it in Markdown since LLMs handle structure and code blocks well. This is where I sketch rough ideas, note tools I want to try, and set expectations. - Ask Claude Code to “ultrathink” on the plan. This runs with Claude’s maximum token budget and produces a more developed version of my draft. -❗CRITICAL❗: Review and refine manually. This step is where ideas turn into real code. I look for assumptions Claude makes, correct missteps, and adjust pseudocode or data models. My developer experience keeps it from going down dead ends—just like pair programming with another engineer. - Run the plan through OpenAI Codex to check for errors. - Have Claude Code implement the plan. - Ask Codex to review the implementation against the plan and highlight bugs.
9 Comments -
Dan Moss
Just updated my portfolio with new robotics projects! 🚀 https://lnkd.in/eitXNuuR My #1 advice for students: Build an online portfolio. Now. Why? It's your visual resume. One glance at your projects can spark more interest than a thousand words. Plus, it helps you reflect on your own accomplishments. It's free, easy, and fun. What is stopping you? #robotics #engineering #portfolio #careeradvice
1 Comment -
Mark Fisher
Composable Agents with Composable Tools It's been a while since I've posted on the Modulewise blog, so I just wanted to share a quick glimpse of what's been keeping me busy. These Agents are configuration only (just metadata for their A2A Agent Cards). All of the integration is provided by reusable Wasm Components that follow a least-privilege capability model and are isolated from each other. Look out for a blog with more detail soon!
3 Comments -
Prashanth Subrahmanyam
Today I decided to experiment with GEMINI.md files, folder hierarchies, system prompts, and overrides on Gemini CLI. I used a few simple prompts and some implicit and explicit contexts in order to understand how Gemini / Gemini CLI thinks and takes decisions to generate code. Ranging from practical exploration, to understanding from the source code. The result - 3 blogs on instruction following within Gemini CLI. Check the comments for the blog links and let me know how you liked this. #Gemini #GeminiCLI
1 Comment -
Kris Jenkins
How do you turn a Raspberry Pi project into production hardware? Can code be as expressive as music? What makes the perfect live-coding environment (and why is it Lisp)? This week we search for the perfect feedback loop for the live-coding experimenter: https://lnkd.in/eStZ6WKH
1 Comment
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More