Ask HN: What interesting problems are you working on? • Hacker News Board

Re: Ask HN: What interesting problems are you working on?

#561

Post by neodypsis » Sat, Sep 17, 2022, 1:51 PM UTC

koeng wrote: post #450

neodypsis wrote: post #44

Earlier quoted context omitted.

Pretty cool. May I ask, what is your background? And, what resources are you using to learn about analog IC chip design?

~10yr of synthetic bio, did 4yr of mitochondrial engineering + directed evolution at UCI, then ran the FreeGenes Project for 3yr at Stanford. Chisel for programming it (great little lang/project) https://www.amazon.com/Digital-Design-Chisel-Martin-Schoeber... Efabless PDKs - https://efabless.com/ Other than that, lots of trial and error. Hardware people are brutal with their acronyms, so it really comes down to a lot…

Very cool. Best of luck!

Re: Ask HN: What interesting problems are you working on?

#562

Post by Ilasky » Sat, Sep 17, 2022, 2:28 PM UTC

FerociousTimes wrote: post #427

Ilasky wrote: post #297

Earlier quoted context omitted.

Task-specific Double communities and interfaces will be the primary difference between Double and competitors. For instance, not all tasks are best, or preferable, when you are on video (e.g., exercising/running). Additionally, we are building Double with the ethos that you don't need to always "crush it", but doing a little bit is okay too. We want to help you build the habits you'd like to see a little step at a ti…

Are these body double volunteers active part of the ADHD community? I can see how this can set you apart from the competition as you have already an active and enthusiastic members among the community doing their best to help each other since they share the same background and life experience.

Double will be matching people with others that are performing similar tasks. In this way, we will be able to foster communities of people that have similar goals.

Down the line, we do have ideas on incorporating “hosting” a Double to help give a sense of service and further reinforce the community aspect as well.

Re: Ask HN: What interesting problems are you working on?

#563

Post by O__________O » Sat, Sep 17, 2022, 2:29 PM UTC

waylandsmithers wrote: post #556

provability wrote: post #401

Earlier quoted context omitted.

I know, right?

Well, welcome to my neuroses- because you asked now it's not as "thirsty" to do that. There was a fantasy sports game in the late 90s and early 2000s called wall street sports, where you'd buy "shares" of professional athletes and the prices would go up or down. My project is my start on bringing some form of it back. [ https://sportfolio.money/ ]

Lots of ventures that have made “investing” in a given real world persona; for example:

https://en.m.wikipedia.org/wiki/Football_Index

https://en.m.wikipedia.org/wiki/BitClout ___

At least for me, your site loads for a split second, then goes completely white; might want to do some cross browser testing, especially on mobile.

Re: Ask HN: What interesting problems are you working on?

#564

Post by curious16 » Sat, Sep 17, 2022, 3:14 PM UTC

Chirono wrote: post #551

curious16 wrote: post #532

Mine is not an interesting "problem" generally speaking. But, to me, it is a problem because I haven't solved it yet. I am on the journey of learning computer science and programming on my own from awesome books and resources available out there. I came across this article by Peter Norvig: https://norvig.com/21-days.html after I read his code of a spell checker: https://norvig.com/spell-correct.html . I was mesmerize…

That code is written in a very declarative, functional style. Haskell is a language that forces you to write code like that, so might be a worth a look if your goal is to write ‘pretty’ code. However, I’d also add that becoming an “elegant” and “useful” programmer are often at odds with each other. It’s very easy to spend so much time trying to make your code pretty with the perfect abstractions that you never actual…

Thanks for the suggestion. Will keep that in mind. I understand that too much perfection can be a roadblock to progress.

Re: Ask HN: What interesting problems are you working on?

#565

Post by O__________O » Sat, Sep 17, 2022, 3:30 PM UTC

techdragon wrote: post #510

An amateur space telescope, I started it as a COVID lockdown hobby project. It’s like a much nerdier “build my own car” sort of project at this point. I’m trying to get it ready for when reusable launch vehicles lower the cost of launching to orbit to a suitable level it won’t bankrupt me. Surprisingly the most difficult thing is communicating to the ground. Not batteries, not imaging, not structural engineering… com…

Curious, what functions specifically fall under ITAR?

For example, have you reviewed something like this page:

https://research.mit.edu/integrity-and-compliance/export-con...

Re: Ask HN: What interesting problems are you working on?

#566

Post by dagar » Sat, Sep 17, 2022, 3:39 PM UTC

eloisius wrote: post #302

Automatic camera calibration refinement/healing. SfM is a well developed area, and there are well understood methods for gathering correspondences like SIFT in order to calibrate cameras. However, it usually requires a separate calibration phase prior to doing whatever you want to do with that calibration. In my case, we’re using it to triangulate 3D objects from 2D object detections done in each of our calibrated ca…

Is this anything you're ready to share? I'm interested in something similar in the context of Drones.

Re: Ask HN: What interesting problems are you working on?

#567

Post by O__________O » Sat, Sep 17, 2022, 3:47 PM UTC

lee101 wrote: post #503

In working on a text/code generation system, similar to OpenAI gpt/codex it's a bit different in that it researches links via crawling and understands any linked images. Also different in that it has reasonable pricing and a free teir. https://text-generator.io/blog/text-generator-now-researches... It also does embeddings for text/images and code. The problem i suppose is general agi where you have unstructured input…

You mentioned pricing as your primary driver, but don’t list any pricing data or price comparisons on your site. Am I missing something?

EDIT: Nevermind. Used a throwaway email to signup and it’s “100 Free requests per month, then $0.01 USD per request” - which unless I am missing something appears to be more expensive that OpenAI’s pricing:

https://openai.com/api/pricing/

Re: Ask HN: What interesting problems are you working on?

#568

Post by techdragon » Sat, Sep 17, 2022, 3:48 PM UTC

O__________O wrote: post #565

techdragon wrote: post #510

An amateur space telescope, I started it as a COVID lockdown hobby project. It’s like a much nerdier “build my own car” sort of project at this point. I’m trying to get it ready for when reusable launch vehicles lower the cost of launching to orbit to a suitable level it won’t bankrupt me. Surprisingly the most difficult thing is communicating to the ground. Not batteries, not imaging, not structural engineering… com…

Curious, what functions specifically fall under ITAR? For example, have you reviewed something like this page: https://research.mit.edu/integrity-and-compliance/export-con...

The largest issues I’ve had with ITAR are:

1: I’m building this myself as an amateur, not a university or company, this seems to instantly put everyone on high alert and I suspect I’m getting “ITAR” as an excellent stonewalling/excuse instead of just saying they don’t want to talk to me.

2: I’m working with an orbital imaging systems which could in theory be used as a very ineffective spy sat (the optics aren’t designed to look at the ground but it is a telescope)

3: I’m not in America, so I have to deal with the “international” part of ITAR. It’s not like I’m in an embargo country, I’m in Australia… but a lot of people want to minimise their ITAR exposure so they just don’t want to work with anyone outside the USA, they are effectively “outsourcing” the ITAR problem to the downstream customers that build systems using their components. Since they aren’t exporting they have a much simpler due diligence process, like just making sure they confirm the buyer is an American citizen.

Overall it’s just been less of a complete dead end but more of a minefield of bullshit I have to dodge any time I’m forced to deal with American companies… dual use technology is ok to buy, until your using it for one of those special purposes, then it’s a mountain of paperwork.

Re: Ask HN: What interesting problems are you working on?

#569

Post by sandreas » Sat, Sep 17, 2022, 3:57 PM UTC

samsquire wrote: post #201

sandreas wrote: post #37

I'm currently working on a file indexer for my pet project tonehub[1] (an audio API like navidrome or audiobookshelf). I thought this is gonna be an easy task, but after running into a lot of problems it turned out that this is indeed an interesting problem to solve. Just to name a few problems I ran into: - Performance: How to scan files as fast as possible? - Tag-Storage: How to store audio tags as generic as possi…

You might find the idea of "cancellation trees" useful. Create a hierarchy of loop data structure of loop iterating variable and loop limit then cancelling any part of the tree cancels its children. If you have a worker thread that has a hot loop that checks a pthread structure as its limit. You can then send a virtual interrupt to it, or preempt it from running further by setting the loop to its limit. You can see m…

Thank you. This is indeed interesting... Currently, I use the `CancellationTokenSource` / `Task` concept of C# and I'm pretty happy with it, but this is definitely worth reading.

Re: Ask HN: What interesting problems are you working on?

#570

Post by sandreas » Sat, Sep 17, 2022, 4:04 PM UTC

minhmeoke wrote: post #534

sandreas wrote: post #37

I'm currently working on a file indexer for my pet project tonehub[1] (an audio API like navidrome or audiobookshelf). I thought this is gonna be an easy task, but after running into a lot of problems it turned out that this is indeed an interesting problem to solve. Just to name a few problems I ran into: - Performance: How to scan files as fast as possible? - Tag-Storage: How to store audio tags as generic as possi…

- Moved files: How to not lose all customized information (rating, playlists, playcount, etc.) when a file gets moved to another place? Many filesystems try to solve the same problem (eg: customize the appearance of files in a particular folder [1]). One solution is adding extended file attributes [2], however this might not be supported on all operating systems. [1] https://en.wikipedia.org/wiki/.DS_Store [2] https:…

Wow this is awesome stuff, thank you very much.

My current solution / plan is:

- Get byte offsets of the audio stream only (e.g. 384, 834882948) - Ignore metadata and build xxhash only over this part - To be faster, build the hash over 5MB in the center - If a new file is indexed, build hash and look it up in the database - If one match is found, assume this is the file - If 0 or more than one, assume new file

The edge case of 2 or more with the same hash has never happened. Worked out pretty well so far, although its not perfect.

And C# File API provides FileWatcher possibility, so I get notified whenever there is a change. The first run has to be done as full scan, after starting the full scan, a FileWatcher is registered and pushes to the Queue whenever a file has changed.

It's far from perfect, but it works.