3d printed rocket engine

Most rocket engines are essentially thousands of parts held together but here there are no welds, bolts, or assemblies required.

August 12, 2025

Happy Tuesday, folks!

Welcome to This Week in Engineering

Last week’s crossword puzzle seems to have been more time-consuming than I imagined. Feedback taken, and we’ll revert to the previous style of games 🤓

We’re now working on building an engineering swag store, and instead of just stickers, you’ll get access to more things. For the winners, we’ll be sending a voucher so you can redeem it for what you like!!

P.S. Is there anything you LOVE as an engineer that we should put on the store? Special legos, lasers, robotics kits? Let me know by replying to this message!

Here’s today’s game:

So I’m scrolling through engineering news yesterday, and I saw this headline about a startup that 3D printed a rocket engine.

My first thought was “cool, another small prototype.”

But, Agnikul Cosmos 3D printed a meter-long rocket engine as one solid piece.

There are no welds, bolts, or assemblies required. Most rocket engines are essentially thousands of parts held together. This thing is one continuous piece of metal.

They use additive manufacturing with Inconel, a superalloy that can handle extremely high temperatures inside rocket engines.

Instead of building layer by layer like normal 3D printing, they've automated the entire process to create complex internal cooling channels, combustion chambers, and nozzle geometry all in one go.

I thought that 3D printing the engine was a big deal. But they got a patent as they cracked how to create internal geometries that would be impossible to machine or assemble traditionally. Think cooling channels that spiral through the engine walls in patterns you couldn't build any other way.

This is huge because no joints means no potential failure spots. Now I might be a bit less scared to sit in a rocket 😂

A good sandwich with some espresso calls for a burp.

But now this is happening with microbes. It almost seems like an AI-generated video…

Jokes aside, Norwegian researchers are now trying to fight climate change using biofilms. These are sheets of microorganisms that work together like a tiny factory, and their structure is what allows for this burping of sorts (ofc not real burps).

Here's the process: They feed carbon dioxide and carbon monoxide to these specialized microbe communities. The microbes use methanogenesis (the same process that creates natural gas in swamps) to break down the CO2 and convert it into biomethane.

As a result of this, we get 96% pure biomethane. That's clean enough to use directly as fuel.

I asked AI to imagine this. And it came up with this:

This, I wouldn’t mind riding in! Kinda like a low-rider cybertruck?

Here's what researchers did: They represented songs as "rhythmic contact chains": precise timing data that tells the robot which drums to hit and when. But they didn’t tell them HOW to hit them.

The robot practiced in simulation, getting rewards for accuracy and timing. Through trial and error, it picked up the strategy we do as humans: switching drumsticks between hands, crossing arms to reach distant drums, and planning movement sequences ahead of time.



The scary part is that these behaviors emerged naturally from the learning algorithm, not from programming. The robot basically figured out that human drummer techniques are actually the most efficient way to play complex rhythms.

It achieved over 90% rhythmic accuracy on everything from jazz to metal, developing its own style.

I used to like seeing humans play the drums with their long hair going crazy all over, but now I guess I have to get used to seeing robots instead 😆

Ever tried to use a flashlight in thick fog?
You point the beam ahead, but all you see is a glowing wall of mist.

Brain tissue is very similar to biological fog. Scientists have been trying to see deep into living brains for decades, and they keep hitting the same wall because light just can't travel through dense tissue and come back with useful information.

So MIT researchers did something that sounds completely backwards: use sound…

What currently happens (old method)?

You shine light into the tissue, and the photons immediately start bouncing around like pinballs in a machine. By the time any fluorescent signal tries to make it back to your detector, it's been scattered so many times that less than one photon in a trillion actually makes it out.

Current brain imaging can see maybe 200 micrometers deep. That's barely scratching the surface. If you want to study important brain regions like the hippocampus (memory formation, learning, and spatial navigation take place here), you're out of luck as it sits more than a millimeter down.

What’s this discovery exactly?

The MIT team, led by Professors Mriganka Sur and Peter So, fires incredibly short laser pulses (20 femtoseconds), which are so brief that light only travels 6 micrometers during each pulse.

These pulses use a trick called "three-photon excitation," where three infrared photons work together to react with molecules that normally need ultraviolet light.

When these photons hit NAD(P)H molecules (these are cellular energy meters which show how hard a cell is working, and they exist in every living brain cell), the molecules heat up by a tiny fraction of a degree, causing the surrounding tissue to expand rapidly.

This creates a pressure wave / a sound wave that travels through brain tissue like a submarine's sonar ping.

Why sound works where light fails

The genius is in physics. Light gets scattered to death in brain tissue, but sound waves get through with barely any interference.

Their ultrasonic detector picks up these pressure waves and converts them back into images.

As a result, they can see individual brain cells through 1.1 millimeters of tissue, which is five times deeper than the best optical methods.

The engineering magic behind it

The technical achievement here is honestly ridiculous. They're creating laser pulses shorter than the time it takes light to cross a single red blood cell, focusing them with precision, and then detecting sound waves that are quieter than a whisper.

The three-photon excitation happens in a focal volume the size of a grain of sand: 2.2 micrometers wide. Only at this exact spot do you get enough photon density for the three-photon magic to work.

And they're doing this 400,000 times per second.

What does this mean for medicine

During brain tumor surgery, surgeons currently face an impossible choice: remove too little and the cancer comes back, remove too much and you might damage critical functions.

This system could provide real-time metabolic maps during surgery, showing exactly which cells are cancerous based on their energy signatures.

Unlike current surgical guidance systems that require injecting fluorescent dyes, this method uses NAD(P)H.

For disease research, this could revolutionize how we study Alzheimer's, epilepsy, and stroke.

The "that shouldn't work but totally does" factor

What makes this breakthrough so satisfying is how it flips conventional wisdom.

For decades, everyone assumed you needed to get photons back out of tissue to create images.

But now we just need to listen to what photons do when they get absorbed.

Sometimes the indirect approach is actually the most direct path to the solution.

What's next

The immediate challenge is moving from lab demos to actual clinical use.

Right now, they need to position the acoustic detector on the opposite side of the tissue, which works fine for brain slices but not so much for living patients.

They're working on same side detection systems that could integrate with surgical microscopes, and they're pushing toward 2 millimeter imaging depth in living brain tissue.

Want to list your job with us?

Hit reply and we’ll get it to over ~20k engineers : )

Reply

or to participate.