Ask HN: What is something exciting you're working on?

56 points by Link- ↗ HN
The more I advance in my career, the more frequent are the phases when all novelty and excitement about software development, and tech in general, starts to fade away.

I'm looking for inspiration, so what's something really exciting that you're working on? (day job, hobby project, whatever...)

103 comments

[ 2.9 ms ] story [ 189 ms ] thread
I’m learning Vue. For both professional and hobby usage.

So far, the switch from jQuery, thrills me.

Welcome! Vue is such a pleasure to work with! Love how you can drop it into a page and add JS interactivity inline OR create large scale component based apps!
Home automation with arduino and Nodered.

I have already implemented the garden lights and the heat on off. I get temp and humidity data from the garden and i want to add some security features and implement a morning routine (warmup the espresso machine, start the heat, open the curtain)

I'm building a little platform to experiment with embodied / real-time multi-modal learning. Basically I have an old boom box that I gutted, stuck a USB power bank and a USB charger in there, along with a Raspberry Pi. The Pi will connect to one or more cameras, one or more microphones, and a handful of other sensors to let it sense its environment to various degrees. The main ones I want to work with are accelerometers so it can sense motion, temperature, and possibly GPS.

The form factor was intended to allow for portability (it has a handle and I can carry it around with ease, and the Pi can run off of the battery when then AC supply is disconnected) and to have enough space to allow for positioning multiple cameras for binocular vision and multiple microphones for stereo "hearing", as well as room for the other random sensors. Oh, and I'm re-using the existing speakers for the audio out so I can work on integrating speech synthesis into the whole kit and kaboodle.

For the software, I'm looking at starting with a super-simple BDI interpreter for the basic "cognitive loop", using an RDF triplestore for semantic knowledge, neural network models where appropriate (object recognition for example), and then start trying to build up from there. Also looking at systems like SOAR, ACT-R, OpenCog, etc. for inspiration.

Of course most of the "heavy lifting" from a computational viewpoint will need to happen on a server somewhere, since a single Raspberry Pi can only do so much. So there's a corresponding server backend piece that will work in conjunction with the remote portion. For now it'll be a fairly low-end server that's physically here at home, but if/when I start needing to scale things up I'll probably switch to using cloud resources on AWS as needed.

All in all, the basic idea is to have an "AI bot" that is alert, observing, and (hopefully) learning all the time and that will learn more like a child learns, compared to the way we train ML models today. That's not to say that there might not be some batch mode training as part of this but I'm hoping to experiment mainly with learning modalities than can happen in real-time. There's no fully fleshed out theory that I'm working off of, but I plan to tinker with a variety of things - Hebbian Learning, Reinforcement Learning, etc. Maybe I'll learn something interesting, maybe not. But it should be fun in either case.

I am making a Lovecraftian text editor called Tentacle Typer. It's a text editor / game where you play as a tentacle monster with a magic mechanical typewriter.

Creative writing releases psychic energy that activates machines / opens doors and all sorts of other things. It also exports .txt files.

Depending on what you write, you'll learn different sigils you can arrange into a circle that will activate when you're in danger. Writing a romance novel gets you different powers than a horror or sci-fi short story, for instance.

https://twitter.com/LeapJosh/status/1459527876118814728

It's pretty weird, but I'm really proud of all that I've done myself this last year. Solo indie gamedev is a perilous, extremely stressful, maybe stupid career choice but I'm realizing a lifelong dream.

I also think I might be inventing/innovating a new genre of game so it's fun to maybe leave a mark that way too. I hope it'll squeeze some creative writing out of people they never would have written otherwise.

Setting up a platform/cloud agnostic PaaS that a scrappy preproduct/early stage/pre-revenue startup can use without incurring large, spiky, unpredictable costs.

Because at this stage the startup has too much going on, this PaaS is very low overhead - so no Kubernetes complexity.

Think EC2/EBS/ELB at flat predictable rates.

(comment deleted)
I've felt strongly for a few years that there is a market (strong interest, if not financial) for a small computer designed to let you be creative and expressive and to learn how computers really work. Over the past year and a half, I've built up a BASIC-like system on my own bare metal kernel on ARM64, giving a boot-to-BASIC programming and prototyping environment in a few seconds. I've worked down the stack as far as building an SBC-based custom hardware with HDMI and keyboard that also works as a handheld portable with onboard display & speaker. You can also do inline ARM64 assembly and memory peeking/disassembly as well as instruction-level step debugging. Polyphonic audio, direct I2C/SPI/GPIO/etc. access (i.e. at the byte level), and more.

It's just radically freeing to not have to wait for a boot, to have direct control over hardware resources like the frame buffer and audio registers, but to start with something as simple as Hello World and go in either direction from there — up in programming complexity or down in systems understanding. I'm very close to getting it into friends' (and their kids') hands. I'd love for it to take off, and have experience actually selling services and products, so I don't think it'll just be a hacker toy, but who knows? Regardless, it's been probably the single best learning project I've ever done. And it's so fun! It'll continue to be so for me whether it sells and/or has a community that builds up around it, or if it's just my passion.

Do you have a newsletter or website? Would love to hear about updates to this.
Not yet, but I'll be certain to post it here when it's in a form where I feel comfortable doing that.
I have had this exact idea before, and agree with you on almost every front except for this seeing widespread adoption by non-hackers. :P

"Radically freeing" is exactly how I'd imagine it too.

If you need any help at all on the non-technical aspects (logistics, fundraising, manufacturing/stock-keeping), please shoot us a message (email in bio). Not asking for anything in return, would just love to see something like this come to market.

So many creators overlook these aspects, and they're arguably more important than the technical specifics.

Thanks, I appreciate this. I have thought of many of these factors, and am well familiar with many of the ways that non-technical aspects of technical projects can surprise/horrify/block/crush you.
Brilliant. In that case, please share your project here when it's ready to go!

I'm sure it would reach the front page very quickly, and I'd love to buy one if only to support you!

I'm researching how we can use technology to improve the way we learn and think.

I'm currently exploring how AI/Natural Language Processing can automatically organize personal knowledge and notes by tagging and linking ideas / content together [0], removing the friction of organization to help you find new connections and ideas.

If you're interested in this domain, I'd love to talk to you!

[0]: https://github.com/Uzay-G/espial

I’m building a hosted data warehouse for different verticals. My goal is to target people interested in doing analysis, but acquiring data and setting up even datasette is too complex, especially if the data needs transformations to be easier to comprehend.

Then I’m building that into a platform where people can fork any query, modify, and publish with their own analysis in order to build a portfolio.

My first market is sports data. There are many aspiring analysts, and I want to 10x the number of people who do this work. And I think the best way to learn Analysis is SQL, and the best way to learn SQL is by building off other peoples queries (learn by example / exploration).

For analytics sometimes SQL is not the best query language. In many practical scenarios analytic queries require looking at most rows of a dataset, but only at a few columns.

This is what OLAP-like engines are built for.

When you have these types of queries, the relational model ends up degenerating in a star schema with queries issuing a join for each data column on the first projection, and then a pass on that projection for aggregation, typically working on a time range that's relatively recent.

For these, native columnar stores are usually a better option. Things like Apache Pinot https://pinot.apache.org/ might be a better fit.

If you add a real-time requirement, it gets even more challenging, going into the realm of custom built query engines, such as those that back products as those built by Medallia or other customer experience companies.

It's a really interesting niche.

Fully agree with everything written here. I've spent way too much time window shopping different database technologies, which are varied and super interesting. For my little hobby project, I'm just using a few gigs of data and am sticking with Postgres to minimize complexity - PG is just such a great tool!

Since I have you here, what do you think about MADlib from Apache? Basic ML in SQL is really appealing - one of my objectives is to teach people analysis using SQL, and it'd be great to only need to develop that one skill (rather than "learn SQL and python"). https://madlib.apache.org/

I am training a reinforcement learning agent to defeat bosses in one of my favorite video games (dark souls 3)

https://youtube.com/channel/UCmoHCPYZviZ0iQ_t_j8cfaw

Holy! I've longn been interested in working on a side-project like this. How do you interface with the game? Do they have exposed APIs? I'd love to know more! Please give us some more details/Github/Bitbucket/blogposts!
(comment deleted)
I am not working on anything specific but I have three ideas around a interesting alternative for google maps/yelp, a dating app that addresses the unbalanced experience of men/women, and a crypto idea that helps reduce your chances of your funds getting seized by the likes of the Canadian govt.

But ideas are cheap, a minimally viable product would be more impressive.

I am building something on top of GPT-3 that translates free form text into SQL commands.
I'm starting to get involved in the space industry (currently an undergrad student), and there are some wild things coming in the next few years. Be ready for people on the moon and Mars, commercial space stations with 50+ people, in-orbit manufacturing, and early asteroid mining by the end of the decade. I'm very excited for the stuff I'm working on, as well as what's coming in general.
This stuff is so exciting, the childhood sci-fi dreams finally starting to come true! Godspeed sir.
I'm trying to solve my addiction of refreshing tabs constantly to get updates from my social media profiles, by building something that does it for me instead

Think rss, but for my personal feeds.

Still mvp, lotsa bugs, but if anyone has similar problems -> https://fetcher.page

I'm working on an online reading group centered around the classics where people can see each other's notes and highlights and start conversations within the book.
I'm trying to create a ML model which can detect if my dog is in pain just by looking at its face. Maybe, I'll create an app for it too if it works well.
How are you planning on collecting data? Seems hard to be a detached trial aid when your dog is in pain…
Currently, I'm approaching this from another angle. Instead of giving training data, I'm trying to use the same inferences vets use to determine if the dog is in pain. For ex: if the dog has flattened ears or grimacing look.

I'm still reading research papers to figure what's the best way to approach this problem. This is just the first step and I'll see what can be done regarding collecting data later on.

When a dog is in pain, it makes a short high pitched noise.

Reference: my dog has problems with his hips and likes to jump... Suddenly the pain kicks in.

Hope it helps

That's pretty cruel, how to you get your test set? :)
You feed him peanut butter and jalapeños
I’ve been working on a job board for startup jobs[0]; if you want to work at a startup, there are more companies than ever, but it’s also harder than ever to know which one to work for. I track number of job posts over time so jobseekers have some data about how much a startup is growing, and I’m hoping to add more types of data as time goes on. Why should investors be the ones who get all the data about startups? Job seekers should, too!

It’s been fun to work on, plus it’s a challenging technical problem: in order to scrape job posts for many companies, you need to make a very generic scraper since they all have different formats and you can’t rely on HTML structure.

[0] https://www.coolstartupjobs.com

I've started upping my woodworking skills. For years I would cut and screw together boards cut from construction grade lumber (cheap 2x4 pine for example). I now work with at minimum Poplar (still a bit on the soft side, but works much like hardwood), Oak, Maple, basically what I can get in the premium lumber section at a big box store, and did some alignment on my compound miter saw. What a difference that makes, have cranked out a hand full of desks and other furniture that is getting better with each piece I made.

Also got into 3D printing, using OpenScad for creating the models. It is almost magical when you can think of something that you want, but they don't make, yet you can make it yourself (different brackets for holding accessories on a bicycle, clips that holds a plexiglass screen in front of the TV to protect it, etc, things like that).

These are a couple of ideas of things that a technical mind can work on (woodworking and 3d printing both require some amount of precision and design). Keeps things interesting.

I'm working on a no-code platform [1]. Given that this space is getting crowded I am starting to investigate adding the ability to create NFTs and make the web3 space more friendly to non-developers which is exciting! (at least to me! :-)

[1] https://www.webase.com

I'm working on a better source-control, starting with game developers.
Derw[0], an ML-family language that targets TypeScript or JavaScript. The aim is to have a language similar to Elm, but can do interop with TS a lot easier. I'm mostly excited that this will make my TypeScript projects a lot more enjoyable for me to work with.

[0] - https://derw-lang.github.io/

Github - https://github.com/eeue56/derw

Announcement post - https://derw.substack.com/p/why-derw-an-elm-like-language-th...

how is this different from Reason?
Roughly, Reason is OCaml, Derw is more like Haskell/Elm. Derw wears the Elm influence proudly, with similar design decisions and goals, and similar standards for tooling. Reason is maybe more appealing to people from JS or OCaml backgrounds, with support for things like JSX. Derw is like Elm: functions and values.
The Haskell is strong with this one!
https://veloren.net/

It's pretty wild that it's possible to make an open source 3D MMORPG where everything is procedurally generated and combat is in real time. That's a combination of words that shouldn't work.

Nice. How about cooperative gameplay? I get the combat appeal, but cooperation ala Minecraft can be truly appealing.
The only real gameplay is raiding and crafting for now. It is possible to build, but you have to be admin and the UX sucks, as it's not intended to be a gameplay mechanic