Top
Best
New

Posted by david927 3/30/2025

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

What are you working on? Any new ideas that you're thinking about?
390 points | 997 commentspage 40
ghoshbishakh 3/31/2025|
I am working on supporting wildcard custom domains for https://pinggy.io

What we found is that we can issue certificates for a wildcard domain even without asking the user to setup TXT records. If one CNAME record is set, we can set the TXT records on our domain to validate the wildcard domain.

Very excited about it.

dpacmittal 4/1/2025||
I've been building an email archiving system [1] which ingests emails sent by a company to it's customers. The hypothesis is that this data is largely un-indexed and once it's indexed, it can be really valuable tool for market insights.

[1]https://pandorasinbox.com

thomasjb 3/31/2025||
I'm putting about an evening a week into Advent of Code 2015, in C, to try to learn a bunch of stuff that I won't unless I push myself: https://git.thomasballantine.com/thomasballantine/Advent_of_...
tschillaci 3/31/2025||
My team’s building micro AI agents that perform very specific tasks really well. Right now, we see ourselves more as a consulting firm, developing custom agents while we search for the right vertical to specialize in. Eventually, we’d love to turn that into a SaaS product. We only have demos to show yet, but I’d be happy to chat more about it :)
sakisv 3/30/2025||
I want to SaaS-ify my scraping process. It's gotten to a point that it works nice and reliably for me, now I need to wrap it in a good DX package and call it version 0.1.

(The main drive behind this was not to sell it, but to have a UI for when a website changes its layout and I'm on holidays and don't have access to my terminal and/or my yubikeys)

attilakun 3/30/2025||
I was trying to reproduce the NeuralSVG paper but it's hard. In the process I had to learn about diffusion and flow matching.
andupotorac 4/8/2025|
Regardless if this or other paper is what you end up implementing, I wanted to say kudos for doing the right thing - having the right approach. Building stuff that's just now possible - in AI, from research papers - and productizing it, is how you get ahead of everyone else.
attilakun 4/15/2025||
Thank you. I got something (but not the full method) working. However, I ended up abandoning this approach because I realized that the SDS process is too slow. It'd take at least 10 mins to generate a single SVG.
yonatan8070 3/31/2025||
I'm working on building my first custom mechanical keyboard, switches (Gateron KS-33 browns), hotswap sockets, and stabilizers should be coming in any day now, just have to do some measurements on the physical stabilizers since the CAD just didn't make sense, then it's time to send off the PCB and print a bunch of keycaps
QuarterDuplex 4/2/2025||
I'm working on revitalizing a FPGA DSP from my first failed startup. It's a RISC-V processor with a vector machine glued onto the side. It's capable of 30 second rapid testing of new algorithms which normally takes hours in FPGA land. Looking for a clear way to describe this novel approach to DSP.
farseer 4/7/2025|
Where can one read more about this?
anshumankmr 3/31/2025||
Working on a tool to make playlists on Spotify using Autogen + Deepseek's R1. https://github.com/anshumankmr/sporky

Got the basic happy path to work, albeit still some tweaks are needed to get it working a bit better and seem more conversational.

ivix 3/31/2025|
Network infrastructure for Bluetooth: https://blecon.net
dansult 4/3/2025|
Thats an interesting project. Thanks for sharing.
More comments...