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

I've been suffering from migraines for the last month, so have channeled my (non-migraine) time into a migraine tracker to try and find the root causes. All the tracking apps I tried all have nice complex forms, which is all well and good, unless...you are having a migraine.

Rough idea is easy to use voice mode to record data, then analyze unstructured data with AI later on.

I want to track all relevant life information, so what I'm eating, meds I'm taking, headache/nausea levels, etc.

Adding records is as easy as pressing record on my apple watch and speaking some kind of information. Uses Deepgram for voice transcription since it's the best transcription API I've found.

Will then send all information through to a LLM for analysis. It has a "chat with your data" page to ask questions and try and draw conclusions.

Main webapp is done, now working on packaging it into an iOS app so I can pull biometrics from Healthkit. Will then look into releasing it, either on github or possibly in the app store. It's admittedly mostly vibe coded, so not sure if it'll be something releasable, but we'll see...

Let me know if this would interest anyone!

Looks great, I’ll check it out!

To prove my expertise in anything from infrastructure, over backend to frontend, I learned how to use terraform to provision a managed kubernetes cluster (on oracle clouds excellent free forever tier).

I am currently developing a web app consisting of a spring/kotlin backend for an angular frontend that is meant to provide a UI for kubectl. It has oAuth login and allows you to store several kubernetes configs, select which one to use and makes it unnecessary to remember all the kubectl commands I can never remember.

It's what I'd like to have if I had to interact with a kubernetes cluster at work. Yes, I know there are several kubernetes UIs already, but remember, this is for 1) learning and 2) following through and completing a project at least somewhat.

One man project for 4 years: https://kastanj.ch/

Most recipes are a failure for beginners on the first try. I aim to make recipes bulletproof so anyone can pick up any recipe and it will just work.

The goal is to make the best recipe app ever. On a technical level recipes are built as graphs and assembled on demand. This makes multilanguage support easy, any recipe can use any unit imaginable, blind people could have custom recipe settings for their needs, search becomes OP, and there is also a wikipedia like database with information that links to all recipes. Because of the graphs; nutritional information, environmental impact, cost etc. can simply be calculated accurately by following linked graphs. Most recipe apps are very targeted to specific geographical regions and languages, this graph system removes a lot of barriers between countries and will also be a blessing to expats. Imagine an American in Europe that wish to use imperial units, english recipes, but with ingredients native to their new homeland. No problem, just follow a different set of nodes and the recipe is created that way for them.

The website is slightly outdated but gives a good idea of what is coming. Current goal is to do beta launch in 2026.

Working on the ArchiveBot dashboard code a bit, on and off. Quite a bit nicer now.

https://wiki.archiveteam.org/index.php/ArchiveBot