I'm working on a motion capture app. The idea is that I can map my movement that's recorded through webcam footage into the movement of a 3d model. This project has involved 3d graphics and AI through 2d pose estimation and 2d to 3d pose lifitng. This project's been pretty fun so far. Here's the github link:
Ask HN: What are you working on? (April 2025)
761–770 of 1001 posts
Re: Ask HN: What are you working on? (April 2025)
#762I’m working on Popgot ( https://popgot.com ), a tool that tracks unit prices (cost per ounce, sheet, pound) across Costco, Walmart, Target, and Amazon. It normalizes confusing listings (“family size”, “mega pack”, etc.) to surface the actual cheapest option for daily essentials. On top of that, it uses a lightweight AI model to read product descriptions and filter based on things like ingredients (e.g., flagging pean…
This is so good I disabled my ad blocker.
Thank you. Seriously.
Note: I searched "Protein bars", and it treated all protein bars equally. The 1st-20th cheapest had <15g of protein per bar. I had to scroll down to the 50th-60th to find protein bars with 20g of protein, which surprised me for being cheaper than Kirkland Signature's protein bars.
Re: Ask HN: What are you working on? (April 2025)
#763I’ve been working on a side project called jobswithgpt — after months of building and refining, the first version is finally live: https://jobswithgpt.com The idea is simple: a job search site that actually works for job seekers. It focuses on listings posted directly by companies (no spam, no middlemen, no bloated sponsored posts drowning out real opportunities). It uses AI to surface better matches, recommend jobs…
love this.
what does like mean? Do i have to just type engineering? Is it scraping data every seconds because applying to jobs just posted is essential
Re: Ask HN: What are you working on? (April 2025)
#764For a more data science-y project, I'm having a blast building https://nhlforecasts.com. I live in a town with a hockey team, and figured the best way to become a fan would be to understand the sport from a data driven approach.
I did a ton of work on building an Elo model first, but was getting very compressed results in terms of postseason predictions. I swapped to a Bayesian approach which has really taken off. Not sure how it's going to handle the second round of games approaching, but that's a problem for the future!
Re: Ask HN: What are you working on? (April 2025)
#765I’m working on Popgot ( https://popgot.com ), a tool that tracks unit prices (cost per ounce, sheet, pound) across Costco, Walmart, Target, and Amazon. It normalizes confusing listings (“family size”, “mega pack”, etc.) to surface the actual cheapest option for daily essentials. On top of that, it uses a lightweight AI model to read product descriptions and filter based on things like ingredients (e.g., flagging pean…
This is so good I disabled my ad blocker. Thank you. Seriously. Note: I searched "Protein bars", and it treated all protein bars equally. The 1st-20th cheapest had <15g of protein per bar. I had to scroll down to the 50th-60th to find protein bars with 20g of protein, which surprised me for being cheaper than Kirkland Signature's protein bars.
My pleasure! Happy you could use it as much as I do. Anyway we can chat in person? I'd love to make more stuff for you. chris@.com
Re: Ask HN: What are you working on? (April 2025)
#766A tree cutting tool. Take photos of the tree from 6 different angles, feed into a 3D model generator, erode the model and generate a 3D graph representation of the tree. The tool suggests which cuts to make and where, given a restricted fall path (e.g. constrained by a neighbors yard on one side). I create the fallen branches in their final state along the fall plane, and create individual correction vectors mapping…
Cool idea. Just wondering why you wouldn't use Lidar for this? I'd have thought the spatial fidelity of a Lidar model would provide a much better model of the weight distribution of a tree.
Re: Ask HN: What are you working on? (April 2025)
#767I just put live a daily logic puzzle, Clues by Sam. I've been working on it for a while now, mainly on the level generator. It was tricky to generate levels that are solvable using logic (no guessing needed), and also fun to solve (no crazy long deduction chains, but also not just obvious things). I'm sure the implementation still has some quirks on some devices, so would love to hear if you encounter issues! https:/…
After getting frustrated by the game not accepting my responses to the obvious clues repeatedly I resorted to clicking on each person and selecting both Innocent and Criminal.
- "My only innocent neighbor is to the left of Harold"
- "Barb and I have one innocent neighbor in common"
- Implies Gary is a criminal
But the game won't let me.
Re: Ask HN: What are you working on? (April 2025)
#768AWS Lambda clone
Re: Ask HN: What are you working on? (April 2025)
#769What a lot of amazing projects! I'm working on a defense drone. I built a garage workshop with a Shapeoko 5 Pro, X1C, soldering station, and learned CAD (ok just fusion). I have a lil drone in the air and I'm adding OpenHD for vtx, Rpi for on-edge compute (Jetson would be better but is expensive). Haven't figured out FHSS or GNSS-denied nav yet (tbh I feel like fhss is gonna be harder). And SITL in a good sim remains…
What is Suna?
Re: Ask HN: What are you working on? (April 2025)
#770Working on a browser extension to make it easier to find content on YouTube that fits your interests. I watch a lot of YouTube videos and have found it very annoying that YouTube latches onto one or two topics that you've watched and only recommends that type of content over and over again. Even if you use their "Not Interested" tool, not a whole lot changes in your recommendations. At the end of last year I launched…
Hm, I tried to contribute, but it asked me to categorize a channel I've never heard of (DefendTheHouse) and no matter how many times I click “skip” it keeps going back to the same one. I also notice that you first said “browser extension” but later you said “Chrome extension”. Are Firefox users going to be out of luck?
Hmm that's strange. Perhaps you have two Google accounts with different subscriptions? When you sign in I grab your subscribed channels and that's what I send when you categorise. You can check the "Subscriptions" page to see exactly what data we pulled.
I did say Chrome browser because with the deprecation of manifest v2, I had to make a choice about which to support. I decided given Chrome's larger market share that it would benefit the most people sooner. However I'm building it in such a way that porting to Firefox shouldn't require much additional work.