Just open-sourced Gust, our task orchestration system. Like Airflow, but faster and much more efficient, written in Elixir. Check it out! github.com/marciok/gust

Am I the only one noticing feature-maxing across open-source projects?? AI can generate features faster than ever, but it feels like every pizzeria is now serving sushi and hamburgers.

AI is probabilistic, not deterministic. 🧠 @MarcioK People are forgetting that for workflows with clear, definitive answers, traditional code is still king. Use AI to build the algorithm, don't let AI be the algorithm—unless you want your deployments to break. #AI #TechTalk

The most underrated language, ecosystem and community 💜

The worst part of doing dev work on a Mac: .DS_STORE

Since the Humanity's Last Hackathon from @huggingface didn’t happen, I set up my own mini version using Kernelbot and Popcorn from @gpu_mode. > The goal was to test how well LLMs can generate code for difficult tasks, like writing faster kernels for Apple’s MPS with @PyTorch. > My strategy was to let the LLM submit a kernel, get feedback from the benchmark, and then iterate based on the learnings. > The hardest part was not the code generation itself, but coordinating all the systems. Kernelbot, Popcorn, submissions, feedback, orchestration... > The benchmark eats almost all my RAM, so parallelizing too many submissions is hard. My machine starts crashing if I push it too much. Overall, I need more time to tune the prompts, experiment with better feedback loops, and maybe try some RL-style iteration. There are still lots of techniques worth exploring here. In the video: Left: task orchestrator Right: live dashboard tracking submissions, code, and lessons learned

Gust reached v0.1.30! Best part: our first major outside contribution 🩵 Install via Igniter: `mix igniter.new my_app --install gust_web --with phx.new` Or run with Docker: github.com/marciok/gust/tree…

Thanks @ElixirConfEU for an amazing time! Great talks and great people. Still unpacking all the feedback and learnings!.

This is the best T-shirt that I’ve seen in a while 😅 @ElixirConfEU

Let's go! I'll be speaking on Friday, 3:20 pm

They arrived! I made a few hats to bring to the @ElixirConfEU If you want one, get a PR merged for any issue tagged `gust-hat` github.com/marciok/gust/issu… #MyElixirStatus #elixir

I fall very easily for book recommendations Especially when they come from highly technical people. Thanks to @0xSero, looking forward to this read.

[#Python vs #Elixir] CPU benchmark: @ApacheAirflow vs Gust - Idle: Gust ~30% more efficient - High load: both hit ~1500% (~15 cores) - Airflow: worker-heavy - Gust: distributed load Much smaller memory footprint with Gust. Full results soon: github.com/marciok/gust

Just finished a RAM benchmark: @ApacheAirflow 3.2 vs Gust 0.1.29. The results are very interesting: - Up to 10x infra savings (idle RAM matters) - ~1.6x more efficient at 50 parallel tasks - ~1.4x at 100 parallel Saving the raw numbers for @ElixirConfEU Full code results coming to GitHub. #MyElixirStatus

direnv is a blessing when working across multiple projects, especially with k8s. github.com/direnv/direnv

As agents get better, we forget when to use deterministic algorithms. Fixed rules, predictable failures -> workflows Open-ended, dynamic problems -> agents The sweet spot: agents coordinating reddit.com/r/AI_Agents/comme…

A human in the loop is still needed—if you want high quality software architecture. My agents most common slops: - Variable and function naming - Repeat code - Don’t write tests before code ;)

Even with a well-defined MCP tool spec, Claude still struggles with type matching. That’s the gap between deterministic expectations and next-token prediction. Still a long path ahead.

Almost done with the Gust MCP server Stateful interactions: MCP Stateless: CLI Like pure functions vs methods.