HNHacker News
TopNewBestAskShowJobs

hannesfur

330 karma · joined March 11, 2024

I am CTO of Stillwind. We make AI tools for Electrical Engineers. I have a background in decentralized systems and networking. Contact me at hannes[@]stillwind.ai
submissionscomments
hannesfur··on How Apple designs a virtual knob (2012)
Having played a lot of MSFS 2020/2024 recently, I feel like I can appreciate this way more now. Since they have to make these knobs realistically and in 3D, when using them with a keyboard and mouse (or even worse a controller) it’s incredibly difficult to see and turn them. It gets even worse since you can push and pull many of these knobs (the difference being potentially catastrophic as well).
hannesfur··on Zoxide: A Better CD Command
Wow! That's really useful!
hannesfur··on Zoxide: A Better CD Command
I also like eza (https://github.com/eza-community/eza) which is a modern ls. Fish is the biggest game changer though.
hannesfur··on Zoxide: A Better CD Command
It definitely got more widespread, but I am not sure that it's related to AI. If it's a way for open source maintainers of awesome tools I use to fund their development, I am totally fine with that.
hannesfur··on Tell the EU: Don't Break Encryption with "Chat Control"
Whenever I look at these proposals I am never sure if the people that wrote that law are not aware that you can’t tap one person without making spying on everyone really easy very quickly, they don’t care or they actually want it. Although this seems like a slightly more sensible version of what they proposed years ago (which was essentially adding the government to every chat).
hannesfur··on AR Fluid Simulation Demo
How much computing resources does it take?
hannesfur··on Ask HN: Fake Video Interviewers
I don’t have a recommendation on how to coach them but i read that some of them are North Korean operatives (https://www.cnn.com/interactive/2025/08/05/world/north-korea...). It’s a real problem and depending on your industry you might be targeted more. In that case a thing that actually works is asking them to say something bad about Kim Yong Un.
hannesfur··on Why Elixir? Common misconceptions
I agree with most commenters that Gleam seems like the thing that’s going to solve a lot of the big problems I have with Elixir.

I didn’t read a lot about Nx and EXLA in the comments, so I thought I’d share my view:

In my opinion, that’s not a strong argument for Elixir, because pretty much every language has libraries like this. Rust, famously, never stops inventing these libraries (https://arewelearningyet.com/), and to illustrate my point, here are some libraries for niche(-ish) languages:

• Crystal (https://github.com/NeuraLegion/shainet)

• Haskell (https://hackage.haskell.org/package/neural)

• Ruby (https://github.com/unagiootoro/ruby-dnn)

• and many more…

They might be suitable for researchers or projects building networks from scratch, but this is already difficult with a library that plays along. For most people, it’s about fine-tuning, combining, and doing inference on existing models (and weights).

People like to criticize PyTorch, but in fairness, with it I can do useful things with recent models (which matters at the speed of things) pretty much the day they have their open source release (thanks to the transformers library). There exists an equivalent to transformers for Elixir (https://github.com/elixir-nx/bumblebee), but it doesn’t support the top 10 trending models on HuggingFace, which is a problem if you want to use them. Qwen 3, for instance, arguably the most useful open source model family at the moment, isn’t supported at all. And this is the issue all of these libraries face: they are not useful for the majority of people needing libraries like this out there.

Also, PyTorch is really well optimized, and there is a good reason that vLLM (yes, it’s not the fastest but pretty fast, supports almost everything, and is the de facto standard for open source LLM inference) is written in Python and PyTorch. Things like Flash Attention 2 are crucial for LLM inference and still hard to find. Yes, OpenXLA is technically very sophisticated, but I never got it working every time I wanted to use it, and JAX seems more comfortable if you want to use TPUs.

It’s still a very strange development that Python runs the AI revolution. Multi-node training might be a lot of fun with Elixir, but here performance (especially making sure the CPU is not stalling the GPU) is vital, which might make BEAM not an ideal choice.

There is hope though. Recently there have been some interesting developments in the Rust world (https://luminalai.com), and of course no HN comment about training and inference is complete without mentioning TinyGrad. But Elixir simply doesn’t make sense to me (for neural networks). I’m sorry.

hannesfur··on Ask HN: What Are You Working On? (June 2025)
Good idea! But try to think of a different name since you will have to deal with Groq (https://groq.com/) and Grok (https://grok.com/).
hannesfur··on Launch HN: Augento (YC W25) – Fine-tune your agents with reinforcement learning
Fireworks has you covered now: https://fireworks.ai/blog/fine-tuning-deepseek-models :)
hannesfur··on Launch HN: Augento (YC W25) – Fine-tune your agents with reinforcement learning
you mean fine tuning that feels like SFT but is different (since you can't use that with reasoning models) built around the DeepSeek class of models?
hannesfur··on Launch HN: Augento (YC W25) – Fine-tune your agents with reinforcement learning
You are right. If you want/can pay out of your own pocket, RunPod (https://www.runpod.io) deserves a shoutout here. We rented GPUs from them (they have them and they are cheaper and more available than Lambda Labs) until we convinced AWS to give us capacity blocks. But in general the prices for GPUs as well as their scarcity is really crass and unlike mining you can't really use gaming or franken cards as a fallback. I can count the GPUs we can do this on (even for relatively small models) on one hand.
hannesfur··on Launch HN: Augento (YC W25) – Fine-tune your agents with reinforcement learning
Yes, you could do that. However, you would have created a different platform than Augento. Maybe we should make the distinction clearer though.

The blog article you are referring to uses another method to fine-tune models that many other big platforms like Together AI (and even OpenAI themselves) are already supporting: Supervised Fine Tuning (SFT). We are doing Reinforcement Learning using GRPO instead. SFT has the big caveat that it requires good prompt-completion datasets to work, which are rare/hard to curate for many use cases. For GRPO, you (the programmer) don’t even need to know what the correct answer is as long as you can decide if it’s a good answer (P?NP) at its heart, essentially.

hannesfur··on Launch HN: Augento (YC W25) – Fine-tune your agents with reinforcement learning
If you send us an email, we can send you your weights right now :)
hannesfur··on Launch HN: Augento (YC W25) – Fine-tune your agents with reinforcement learning
Not yet, but since people seem to want this it’s at the top of our roadmap. If you train a model now and want to get the weights, just message us and we‘ll give them to you ;)
hannesfur··on Launch HN: Augento (YC W25) – Fine-tune your agents with reinforcement learning
That’s fair! It has been mentioned before so we‘ll likely build that into the platform. Would you like us to upload your model to your huggingface account, download the weights or choose an inference provider we then upload it to?
hannesfur··on Launch HN: Augento (YC W25) – Fine-tune your agents with reinforcement learning
In a sense: You are not wrong! But when we got started we thought it is way easier than it actually was. Procuring powerful GPUs alone is difficult, collecting proper data too. But of course you can still do everything yourself. If you want to give this a try yourself, I would recommend taking a look at torchtune (https://github.com/pytorch/torchtune).
hannesfur··on Show HN: Decentralized robots (and things) orchestration system
We chose batman-adv because it's the most widely adopted and supported implementation of a wireless ad hoc mesh network :) Freifunk is running deployments with several thousand nodes.[1] Since that was the scale we were aiming for, this is what we chose. We also considered 802.11s for a long time, but the hardware support is quite poor, especially on a device where people would try something like this: Raspberry Pis.

You are right that mDNS is a popular choice, and we even have it in our codebase to be able to debug on macOS.

However, mDNS is a flooding/multicast protocol, and although batman-adv optimizes those, it puts unnecessary strain on a mesh network. That’s why we didn't go with it.

It also gave us the opportunity to try out this other, in our view, more elegant approach to discovery.

Since IBSS is not really useful outside of Linux popular device support, like smartphones, it fell out of the picture pretty quickly (sadly). However, you can connect bridges to our mesh network (again exposing a batman-adv feature), which would give you the ability to connect phones (and of course, printers) to your robots.

[1]: http://grafana.freifunk-muensterland.de/goto/lAqihJDHg?orgId...

hannesfur··on Show HN: Decentralized robots (and things) orchestration system
When you setup the robots you could load them with the PKI and then load each other robot joining with a signed certificate. Not ideal, I admit.

Another way would be to somehow prove that you belong.

hannesfur··on Show HN: Decentralized robots (and things) orchestration system
That's exiting! I would love to talk to you more if you're open to it :)
hannesfur··on Show HN: Decentralized robots (and things) orchestration system
Possibly yes, I can't give you a number on how quick batman discovers neighbors (will look into that) but we poll it every second. For this use case it could work!
hannesfur··on Show HN: Decentralized robots (and things) orchestration system
Like others suggested a basic step would be to use a certificate based approach where a company (or basically any deployment) gives out certificates for robots allowed to join and you only communicate with them.
hannesfur··on Show HN: Decentralized robots (and things) orchestration system
That's actually a pretty great idea! I will look into it! Thanks!
hannesfur··on Show HN: Decentralized robots (and things) orchestration system
We do discovery via the mesh, yes, but instead of broadcasting (like mDNS), we query batman-adv for the current visible neighbors. If a new neighbor is discovered, we will contact them directly (via WiFi) to exchange the addresses in the P2P network and then dial them. From that, we populate the local Kademlia routing table with the content of the neighbor.

Security is still a big issue. In the current state, there is no security other than application-layer encryption (QUIC & TLS v1.3). That is fine for pilot projects, but it should not be used for anything sensitive. Maybe we should point this out more clearly in the docs. However, some Wi-Fi chips (not the ones on Raspberry Pi, sadly) also allow setting a password in adhoc (IBSS) mode and 802.11s has native support for encryption. In general is here a problem with lack of adoption of standards by the WiFi chip manufacturers and with Broadcom (the chip on the RP) a lack of support in the Linux kernel driver.

We are planning to implement authentication and encryption in the upcoming release, but this might be a paid feature.

Typically, enterprise networks are encrypted via 802.11x (since a leak of the key wouldn't compromise the whole network), and we might be able to build a decentralised Radius server, but I'm not very fond of that idea.

Ideally, the damage one can do by joining the network unauthorized should be very limited anyway, and authentication and encryption should happen on Layer 5.

Would love feedback / inspiration / suggestions

hannesfur··on Show HN: Decentralized robots (and things) orchestration system
You are right. At the moment, we are an opinionated wrapper, but we take a different approach to discovery than other libp2p-based networks with our custom batman-adv-based neighbor discovery.

Abstractions for collaboration are currently in the works, and we hope to release that soon. The work on consensus has already started. Your suggestions seem all very interesting, and we'll definitely consider them. We are also currently in the process of talking to potential users to build handy and approachable abstractions for them.

I saw that [freenet](https://docs.freenet.org/components/contracts.html) went with CRDTs, but I think they made it too complicated. We were thinking about a graph (or wide-column) with an engine similar to Kassandara and a frontend like (or ideally just) SurrealDB.

I remember that iroh moved away from libp2p when they dropped IPFS compatibility and moved to a self-built stack: https://www.iroh.computer/blog/a-new-direction-for-iroh When we got started, the capabilities of iroh didn't really fit our bill, but it seems like it's time to reevaluate that. As a former contributor to rust-libp2p, I never quite got the frustration with libp2p that many people have, Iroh included, especially since many of the described problems seemed fixable, and I would have preferred if they did that instead, and libp2p remains the shared base people build these things on.

I remember Actyx being a rust-libp2p user, but I wasn't aware that they failed. Do you have more info? How and why? It would be great if we could learn from them.

Grammar will be reviewed ;) thank you!

hannesfur··on Show HN: Decentralized robots (and things) orchestration system
Awesome! From what I see, the clock synchronization can be implemented with our SDKs (mainly pub-sub).

I think the size estimation could also be implemented within the provided abstractions (mainly request-response) but might require you to keep track of neighbors. I think you could implement both algorithms by using our SDKs (none for Go yet).

If you need more control or performance, beyond what we expose through our SDKs, you might need to write a custom libp2p behavior and add it to our daemon. The libp2p part is fairly involved, but I would love to help you with that. Either way I would love to help you out :)

I'm so disappointed that I've never seen your blog before. The stuff you write about is so interesting and actually addresses some issues we are facing. I just sent you an email :)

hannesfur··on Show HN: Decentralized robots (and things) orchestration system
Actually, just very briefly at a friend’s ;)

Thank you! So far, we have tested it with Raspberry Pi 4/5. Jetson boards are on backorder. We have some Intel WiFi chips (since they support some stuff we want), and we will get around to trying them next.

The binaries were also tested on x86 machinery.

In general, I'm not too worried about hardware support since batman-adv is quite widely deployed on a diverse set of hardware and the rest is hardware agnostic.

hannesfur··on Show HN: Decentralized robots (and things) orchestration system
Please do and let us know if you have any questions!
← PreviousPage 2 of 2