I did a lot of leetcode stuff this year. Dozens of interviews, no takers.
What got me a job was that I was a solo founder running a business and learning how to make the most of limited resources. Aka I was someone that gets things done and spends the least amount possible to get there. Use the resources you have and show you can create value.
Again, it all comes down to show don't tell.
Leetcode is valuable as a way to practice and maybe reaffirm skills. It's not useful for hiring in a direct way.
This breaks my heart a bit. My first browser extension Cacheout was around 2005. Back in the days of sites getting hugged to death from Slashdot. The extension gave right context menu options to try loading a cached version of the dead site. Tried Google cache first, then an another cdn caching service I can't remember, and finally waybackmachine. Extension even got included in a cd packaged with MacWorld magazine at one point.
This has always been one of Google's best features. Really sad they killed it.
> Tried Google cache first, then an another cdn caching service I can't remember, and finally waybackmachine.
Coral Cache maybe? The caches you listed were my manual order to check when a link was Slashdotted.
Google's cache, at least in the early days, was super useful in the cache link for a search result highlighted your search terms. It was often more helpful to hit the cached link than the actual link since 1) it was more likely to be available and 2) had your search terms readily apparent.
Wow. I'm surprised this app has been in development for so long.
I used Finale in highschool in the mid to late 90s to write (admittedly terrible) music when I was in high school. Really amazing program at the time and the instrument voices were good for the time.
I'm sure I have an old piece somewhere floating around on a floppy from that time in a box.
Does anyone happen to know how far back the latest version supports their old file formats?
I'm not a professional user of Blender, but it's my only tool for working with and creating models for 3D printing. Never could develop a liking for Fusion360, though I recognize the advantages that CAD programs give.
My biggest tip and gripe as a Mac user is to get a decent mouse button remapped utility and combine that with creating a Mac specific settings save file. For whatever reason, MacOS to this day doesn't recognize the extra mouse buttons that Windows does. Like Button4 etc.
Even today, "middle mouse" button is one of the weirdest default choices Blender could make. The middle finger of either hand is an assisting finger, not primary. The index and ring fingers are stronger and it's weird to assign actions that require persistent pressing to a finger that has less strength than the other fingers.
With a multi button mouse you can remap the middle button to the thumb. It's infinitely a better and more intuitive choice.
Also, Blender's trackpad support on MacOS is horrendous, despite the amazing advances we keep getting everywhere else with each new release.
I would love to see a new feature that is a voluntary checkbox for Mac users that enables a key binding layout dedicated to us.
I like the idea of using blender, but I'm under the impression it wouldn't work well for the kind of things I work on, which require precise, adjustable measurements, and parametric design. (Like, custom hardware to mount thing A onto thing B)
Blender is mainly for visualisation, but I've used it for 3D printing. You can set up models in a parametric way by using modifier stacks (e.g. screw, solidify, subdiv etc) or geometry nodes, but I imagine other software is more advanced in this regard, or needs less customization and thought.
Do check the settings! IIRC blender specifically has a toggle between a 2-button and 3-button mouse. Maybe even single button is possible, I remember using ctrl and alt with a stylus.
I'm a full stack eng that recently transitioned to a full time backend role.
I'd suggestion learning about pgvector and text embedding models. It seems overwhelming at first but in reality the basic concepts are pretty easy to grok.
pgvector is a Postgres extension, so you get to work with a good traditional database plus vector database capabilities.
Text embeddings are easy to work with. Lots of models if you want to do it locally or adhoc, or just use OpenAI or GCP's api if you don't want to worry about it.
This combo is also compatible with multiple vendors, so it's a good onboarding experience to scale in to.
Exactly. Marketing dollars come from investors. A big part of modern marketing for startups is spending tons of investor money to "grow" quickly. How long the fire burns doesn't matter, it's all about throwing dried out February Christmas trees onto the flames for quick bursts of flame.
As far as I know (and its not very far to be fair), the main way to generate hydrogen would be via electrolysis in water, which requires electricity. Which means you have to turn that heat into electricity first anyway. This is one of the big critiques of using hydrogen as an everyday fuel source. Takes more energy to produce that you can recover. Its super hard to store (requires very cold temps, or high pressure), and it has a habit of wiggling through any other material's atomic bonds and overall weakening the container material.
But if there is another chemical method that used thermal energy in the process to produce hydrogen then there might be some possibilities in the idea.
Helion https://www.helionenergy.com/technology/ is a commercial fusion company working on a design that theoretically would use direct energy capture from the magnetic fields generated during the fusion event. They made some headlines last year. Not clear if their approach will be successful but it certainly is an interesting approach.
Can someone who has a firm grasp on this stuff explain to me how nuclear reactions don’t create massive amounts of electromagnetism that we can just capture directly? Is it really just heat that it produces?
It generates high-speed neutrons, helium cores, photons, fissile fragments, and in rare circumstances free protons.
Of those, only the protons have an electric charge you could use in any form of generator… but I’m not aware of any form of reaction that predominantly creates protium. The fragments are also charged, but the implication that you’re using fission means they’re in the middle of a block of uranium and won’t keep their speed for long.
The others ignore electromagnetic fields for the most part, and will fly around until they smash into something and go goooong like the world’s smallest bell. Or smash through something, perhaps; it’s a chain reaction after all.
This mostly just makes other stuff move about. Repeat a few dozen times, and you’ve got heat.
The helium nuclei a.k.a. alpha particles are also charged, like the free protons.
However, the alpha decay of some product of the fission reactions does not change the total charge of the fissile material, because the emission of a helium nucleus with a double positive elementary charge leaves a heavy nucleus with a diminished nuclear charge, by those two elementary charges.
Only when the alpha decay happens to occur close to the surface of the material, the positive helium nuclei may escape from it and they could land on a collecting electrode, making that electrode positively charged and leaving the fissile core negatively charged. However such a process would extract only a negligible part of the energy produced by fission. Even if the alpha extraction could be enhanced somehow, the decay energy of the fission products is small in comparison with the energy produced by the initial fission of the uranium nuclei. Extracting directly from the fissile material the nuclei generated by fission would be much more difficult than extracting the helium nuclei.
The Darpa project may succeed to stimulate the creation of some electric generators that could deliver additional energy from a fission reactor, by direct electric charge separation, but that would remain a small part of the total fission energy, most of which will still have to be extracted through thermal methods, like today.
They kinda do. When a fissile atomic nucleus splits, the daughter nuclei repel each other; they are both positively charged and the strong force is no longer holding them together. So they fly off in opposite directions at a measurable fraction of the speed of light. But they don’t generally get very far, because they are embedded in a solid fuel pellet. They can push their way through a few µm of uranium before they are stopped, bumping into thousands of atoms along the way. That’s really where the heat comes from; all that electrostatic force accelerates them to high velocity, but they dump it all very quickly into the material around them as heat.
I mean, that's just because water is a simple medium we can just literally trow out in the environment after use, using basically anything else to spin said turbines would require a closed system and cooling.
Gravity batteries could be a thing but pumped water is the best version of that system, in this case we don't boil the water tho
Water has really awesome phase change properties and is nearly ideal for this kind of situation too. Only thing better is potentially super critical co2.
There definitely trendy since some high profile YouTubers were invited to tour! I sincerely hope they can succeed. I believe a limiting factor is going to be fuel unfortunately, as well as I’m not entirely sure it’ll be radiation free.
The original fuel is deuterium, which is absurdly abundant on Earth. Pure deuterium fusion results in half helium-3, and half tritium which decays to helium-3 with a 12 year half-life. Then their main reaction is D-He3.
The D-D reaction is less energetic and produces a neutron, and the D-He3 reaction doesn't produce a neutron. The combined reaction would release about five percent of its total energy as neutron radiation. The neutrons from D-D are about as energetic as fission neutrons, rather than the extremely high energy of D-T fusion neutrons.
They'll never run out of fuel; there's enough deuterium in your morning shower to provide all your energy needs for a year. But the need to breed He3 will put a limit on how fast they can possibly scale up. It could be that manufacturing will be slower than that anyway though.
Fusion is infinitely harder for this than fission. No company has demonstrated stable fusion with a positive net energy gain. Most of these startups are borderline scams for milking gullible VCs. Helion in particular has been around for more than a decade and was supposed to reach break even in 2023. They haven't even achieved a fully stable D-D reaction so far. The biggest thing it has achieved is siphoning tons of money from OpenAI's investors because of some questionable actions by Sam Altman.
All of those experiments say they are "on schedule" for whatever schedule they made up for this year. If you have been following this for as long as I have, you'll know that these statements are worth nothing.
You're the one who referenced their earlier published schedule, which said they were "supposed to reach break even in 2023." They had a couple years of delay in funding after making that statement.
I think in a Deutritium-Tritium fusion reaction a lot the energy is in the neutron. And since neutrons are neutral you can't really directly convert that into electrical energy.
Yep, for D-T 80% of the energy is in neutrons. Helion is using D-D/D-He3, and for that it's about 5% in neutrons, and most of the rest in fast-moving charged particles.
So they have a simple way to extract electricity directly. They squeeze the plasma with a magnetic field from a copper coil, then there's an explosion of charged particles, which pushes back against the magnetic field and causes electricity to flow in the coil.
As I said, the combined reaction generates 5% of its energy as neutron radiation. The reason it's only 5% is that D-He3 is a much more energetic reaction.
The main issue with D-He3 is that it's more difficult.
I have seen this whole scheme play out more than once by now. Helion is not the first fusion startup. Not by a long shot. It's just one of those who gathered media attention (and even that was not due to their science but to OpenAI's business practices). If you're an investor, you're welcome to hire me as an adviser and I'll tell you exactly where and how they are being delusional or outright dishonest. But most investors get hitched on nice pitch decks and certain keywords or names.
I don't remember the details, but the last time I looked into Helion I came away with the belief that their technology flat out doesn't make sense and will likely never be anywhere near net-positive. Like, the numbers literally don't add up and their design could never be anywhere near net-positive.
I had the impression that the energy flux required to break even is higher than any known materials can support, at any geometry. I really hope that's wrong :-)
They claim to have performed fusion, but not yet produced net energy. So they're very much a fusion company, but not an energy company:
> In 2023, we will end operations on Trenta, our 6th fusion prototype… Our results suggest that Trenta is currently the best performing privately-funded fusion machine in the world. After these last weeks of plasma operations under vacuum, we will retire Trenta and move all focus to Polaris, our 7th fusion prototype, expected to demonstrate net electricity in 2024.
So, they have 143 days left to make good on their current timeline, I guess.
They're an energy company that has failed to produce any yet because of how they chose to generate that energy. Unfortunately, like everyone else who tried this so far, they can make tiny bang go flash, and that's unfortunately still just about it.
So, indeed, so far they failed at energy, and they've only succeeded at fusion in the same way and at the same scale that university labs have.
So I guess you're right: they're not even any kind of company at the moment. They don't sell anything. They're an R&D lab.
What got me a job was that I was a solo founder running a business and learning how to make the most of limited resources. Aka I was someone that gets things done and spends the least amount possible to get there. Use the resources you have and show you can create value.
Again, it all comes down to show don't tell.
Leetcode is valuable as a way to practice and maybe reaffirm skills. It's not useful for hiring in a direct way.
reply