Live data from Hacker News

Ask HN: What are you working on? (May 2025)

news.ycombinator.com

391–400 of 1001 posts

Re: Ask HN: What are you working on? (May 2025)

#391

I recently quit my salary job after 16 years and am consulting in nuclear engineering now. I have a few passion projects that I'm working on (between the somewhat substantial consulting work that came out of the woodwork): - Nuclear Reactor Starter Kit --- an open source set of procedures, processes, templates, and maybe even some IT advice that should help newcomers start companies with nuclear quality assurance pro…

Just curious: as a SWE, what are the prospects like in this field? What kind of background would be necessary to step in and what does the demand look like?

I've always been fascinated by nuclear and it seems like a field that's never going to be not needed, but I kind of suck at physics and chemistry.

Very cool stuff to see on your end, nonetheless!

Re: Ask HN: What are you working on? (May 2025)

#392
post #106

I am working on a music recommendation algorithm for your self hosted music. Think of it like your personal pandora. Backend is already working: Boldaric https://github.com/line72/boldaric And a simple iOS native front end (which I haven’t submitted to the App Store yet). Tor Jolan https://github.com/line72/torjolan It has been interesting tweaking the algorithms and models for various similarity searches. I really l…

Wow! I was thinking about doing something similar (extracting audio features to find really similar songs) some weeks ago. I love the idea, it'd be great to have something like this to discover music too, but it'll be difficult to discover new music because it won't be present in the collection.

I think for exploring new music last.fm or listenbrainz are still good options.

Listenbrainz has some cool tools like troi to come up with weekly lists.

They work on popularity and neighbors, so if you listen to a lot of underground music that isn’t on major platforms, they fail

Re: Ask HN: What are you working on? (May 2025)

#393
We’ve been working on a lightweight tool to help small teams and individuals manage and publish content across multiple social platforms — currently focusing on Facebook, Instagram, and TikTok.

Most tools we’ve tried are either too bloated, too expensive, or limit you to one or two accounts per platform. We’re aiming for something simpler: connect multiple accounts (free up to 3 per platform), schedule posts, organize media, and view basic performance stats — no AI or calendar features yet.

Still early, but the core workflow is live and usable. Would love to hear how others here handle multi-platform content publishing, especially if you’re doing it without a full-time team.

Re: Ask HN: What are you working on? (May 2025)

#394
An agent based simulator with LLMs. When I did my PhD in 2016 I used Agent Based Simulations to predict how to influence humans with chat agents on a large scale. The main criticisms with such simulations is always the realism of the agent (and network).

Now, with LLMs, this got much simpler. As LLMs are really good at mimicking humans.

The first application will be to simulate a LinkedIn post against your own (virtual) followers. I grab your actual LinkedIn followers, "turn" each one into an LLM and see how they react.

Re: Ask HN: What are you working on? (May 2025)

#397
Converting my home server from Docker + Docker Compose to Podman + systemd. (The server runs self-hosting for my movies, TV, ebooks, comics and photos, networked to my other devices via Tailscale).

The DevEx isn't quite as easy, but I'm very happy with the final result. I don't have to run the overhead of a Docker Engine anymore, the network policy is far simpler to write and audit, and I can use the same logging tools for both containerized and non-containerized services.

The main hurdle has been file permissions for the mounted volumes for my data; Podman is rootless by default, which means you _should_ build your containers to run as non-root UIDs/GIDs and map them onto host UIDs and GIDs, then grant permissions for those host users and groups to access your data volumes. In practice, the easiest path is usually to run Podman in rootful mode, which is not a best security practice but avoids the difficult-to-troubleshoot file permission errors if you don't do the UID/GID mapping correctly.

However, unless you are really trying to optimize for overhead/performance, you should probably use k3s.

Re: Ask HN: What are you working on? (May 2025)

#398
post #259

Ive almost finished the first version of a desktop video library app I've been writing for myself. I had the idea last year, but the cost of sending images to an LLM made it too expensive (to run over about 1500 videos), but now it's fairly reasonable. In the app you pick a folder with videos in it and it stores the path, metadata, extracts frames as images, uses a local whisper model to transcribe the audio into sub…

Neat! Have a repo?

Re: Ask HN: What are you working on? (May 2025)

#399
post #113

A new YouTube app/player, for my kids. It allows us to control the algorithm. It’s all LLM translating to YouTube search queries under the hood. Visually it looks the same. The suggested videos come from predefined buckets on topics they love. E.g. 33% fun math, 33% DIY engineering, 33% creative activities. Video recommendations that have a banned word in the title/desc don't get displayed e.g. MrBeast, anything with…

Very interested and will reach out, thanks!

Re: Ask HN: What are you working on? (May 2025)

#400

I've been collecting and digitizing vintage print advertisements and publishing them ( https://adretro.com ). I have tens of thousands of ads in the collection and it would take me many lifetimes to complete, but I've been using AI to extract and catalog the meta data. I can get through about 100 ads/day this way. One of my favorite ads, a computer from 1968 that "answers riddles": https://adretro.com/ads/1968-digi-c…

This is absolutely fantastic! What a treasure trove you're building. I love that you're using AI to help with the metadata extraction while still requiring the physical magazines, that's such a thoughtful approach to preserving these cultural artifacts. That 1968 computer ad that "answers riddles" is pure gold. Can't wait to see more gems as your collection grows!
Post reply on HN