Some ideas are born at a conference between two talks. Others strike in the shower. And some come to life at 2am, in front of a terminal full of logs and a half-working makefile. This year, instead of the traditional eBPF Summit, we tried something different: a global community hackathon, open to everyone, with one simple goal - build something with eBPF, Cilium, Tetragon, or Hubble. And guess what? Magic happened đȘ As well as announcing the winning entries, weâll walk through what made this hackathon special â who showed up, what they built, and why it matters for the eBPF ecosystem. By the Numbers Over the course of four weeks, more than 500 participants from around the world joined the challenge. Some were already deep into kernel tracing and XDP packet manipulation. Others were writing their very first lines of eBPF code, documenting their journey as they went. We saw policy visualizers, schedulers, syscall tracers, CLI tools, tutorials, learning platforms, PRs to upstream projects â and some projects that made us say, âwait⊠thatâs really interesting?!â 549 participants from all over the world and from different backgrounds 42 submissions per track: eBPF Starter Track, Using eBPF and Cilium Technologies 11 volunteer judges from across the experts of eBPF, Cilium technologies and OSS. What the participants says about the hackathon: Great experience! The hackathon motivated me to build something real instead of just reading docs. Suggestion: more beginner-friendly examples for uprobe/tracepoint would help newcomers get started faster. All good, I love the idea this year to do something practical. I would never have done this project otherwise. đ„ And the Winners are... đ„ Without further ado, letâs announce the winning entries from the three Hackathon categories: eBPF Starter Track for Beginners SCX_MUS - Mostly Unfair Scheduler: A custom Linux scheduler that uses the sched_ext framework to prioritise containers in Kubernetes. The project provides a simple CLI that selects a container, extracts its cgroup ID and stores it in a BPF map. Judges highlighted how impressive it is to see a group of newcomers tackle such a complex program type: âsched_ext is one of the more complicated program types to write ⊠itâs a great example of eBPF superâpowers.â Using eBPF Track xgotop: Realtime Go Runtime Visualizer A lean but powerful profiler that instruments the Go runtime to visualise goroutines and memory allocations in real time. With subâmicrosecond overhead and a polished UI, xgotop was dubbed âpprof on steroidsâ by one judge, who also praised the clear architecture diagrams and demo. Cilium Technologies Track Goya: A proofâofâconcept framework that blends the Rustâbased Aya framework for kernel space with Ciliumâs Go libraries for user space. It offers a template for developers who want to write eBPF kernel code in Rust, while still benefitting from the ergonomics of Go for their user space code. Judges appreciated the niche appeal: âIf anyone ever wanted to do Go + Rust, this is a really good starting point.â Special Shout out The judges also decided to award a special additional prize to PythonBPF. As a pre-existing project it didnât strictly meet the Hackathon criteria, but we wanted to recognise it for its potential impact on the eBPF ecosystem. It allows developers to author BPF programs entirely in Python. It includes linting, typeâchecking and a userspace library (pylibbpf), making prototyping far easier for Pythonistas. Judges noted that it âcould unlock eBPF for a whole new audience.â Together, these winning projects show how diverse and inventive the eBPF community has become from scheduling experiments and new language bindings to polished profilers and hybrid frameworks. Theyâre proof that when you give people a kernel hook and a challenge, you never know what magic theyâll create. What the Community Built Choosing the winners wasn't easy, given the high standard of entries. Hereâs a taste of some of the other standout submissions. đ Learning & Experimentation A whole cohort of entrants focused on sharing what they learned. Platforms like User-space eBPF VM (Interpreter) and Debugger and BlazeBpf offered bilingual tutorials and offline sandboxes to help newcomers write their first programs. Judges appreciated how these projects opened the door for beginners, with one commenting that âlearning platforms are the onâramp that makes BPF less intimidating.â đ CLI Tools & LowâLevel Utilities Many participants built commandâline tools and libraries for realâtime tracing, profiling and debugging. There were syscall tracers, profilers, and utilities that visualise container lifecycles at a glance. đ„ Interfaces & UI Innovation Not everything lived in a terminal. Several projects showcased UIs that transform Cilium/eBPF data into dashboards and visual simulations. From Cilium PolicyPilot to DNSâSentinel, teams created web apps that generate and test network policies, visualise flow logs, or even monitor heart rates via Hubble. One of the judges applauded the policy simulator for making Ciliumâs rules âfinally understandable to nonâexperts.â Why It Matters and Whatâs Next This hackathon was more than a collection of neat demos; it offered a glimpse into where eBPF is heading and why those developments matter. CPU Scheduling On the technical front, CPU scheduling emerged as one of the most exciting frontiers. With the Linux 6.12 release, the kernel introduced sched_ext, an âextensible schedulerâ class that lets developers implement and dynamically load custom CPU schedulers as BPF programs (free5gc.org). Unlike traditional scheduler tweaks, which require kernel recompilation, sched_ext exposes a set of hooks (via BPF struct_ops) that allow a scheduler to decide how tasks are enqueued, dequeued and dispatched. The projectâs collection of schedulers (like scx_simple, scx_nest and scx_bpfland) shows how eBPF can tailor CPU scheduling for different workloads. While the ecosystem is still young and documentation scarce, experiments like SCX_MUS or Schedra projects highlight just how accessible and powerful this new capability can be. Which language do you speak? Language choice was another recurring theme. For many cloudânative engineers, Go remains the pragmatic choice: Goâbased libraries like cilium/ebpf wrap eBPF syscalls in a clean API and benefit from a vibrant community. However, Python and Rust are increasingly present. PythonBPF aims to make kernel programming accessible to Python developers, while frameworks like Aya and the Goya project demonstrate that Rust offers both safety and performance. The variety of languages reflects the broader community: thereâs no single âright wayâ to build with eBPF â the point is to choose the tools that fit your background and project. đș Watch eCHO Stream Hackathon Recap with eBPF experts and judges! In the eCHO livestream, we will demo the winning projects and discuss eBPF emerging trends, framework and tooling! Conclusion Donia - On a personal note, I was thrilled to see so much intelligence, curiosity and diversity on display. The creativity was infectious - from firstâtime contributors writing their inaugural BPF program to seasoned engineers pushing the boundaries of what eBPF can do. At the same time, we know thereâs more work ahead to make the community even more diverse and welcoming, and weâre committed to that journey. Finally, thank you to everyone who participated and supported the hackathon, from the judges to the organizers to the participants themselves. Your energy and contributions made this event a success!