
26, Ex-SWE → Solopreneur 📈Working on #Canvograph ($0 MRR) 🗺️Submitted #Luxiom for Appstore ($0 MRR) ✉️Blocked @Minmailist due to CASA
Terminal
Joined October 2025
- Tweets5.5K
- Following518
- Followers679
- Likes9.8K
Pinned Tweet
Finally reached and crossed 600! Thank you for your support everyone, love y'all🤝
This month is gonna be a blast, trust me. I'll be sharing and documenting Luxiom and Canvograph LIVE. Every step, and best part - I'm a noob in distribution and this is where we all struggle, let's be real😂
Is there an issue with chatgpt web's PDF upload?
It's taking sooo long
And my doc is just 130 kb lol
DevDiary retweeted
Today, I get a chance to be THAT senior who everyone turned to for help at their college.
The Lecture on Embedded lInux that I am supposed to deliver starts in 90 mins lads, and I am nervous as fuck when it comes to this. Because this is a chance for me to introduce my students to concepts and topics that their college keeps telling them are "way too advanced" for them to learn.
After the general pleasantries and the introduction about linux and how it's actually NOT an OS and is just a kernel; and the customary differences between Desktop Linux and Embedded Linux, here is what I plan to talk about.
I am walking then through the journey of booting Linux on a Single Board Computer (SBC). The board of choice - a BeagleBone Black. Based around a Texas Instrument's AM335x Processing unit, it follows a 3-stage booting process.
Stage 1 has the BootROM, SPL, and the MLO.
The BootROM is what runs when you power on the board. It brings the board to a point, where it effectively scans 3 different address points from where the next stage of the booting process shall arrive from. Those can be USB, SDCard, the on-board eMMC, or even network booting too!
In secure boots, it also scans the incoming images and performs authentication on them BEFORE letting it run.
Then, the SPL (Secondary Program Loader). This piece runs on the SRAM of the processor, and it's only job is to ensure that the much bigger DRAM (or DDR) is initialized, and ready to pick up the next batch of images called the MLO. This is necessary because the SRAM may not be big enough for the incoming images, and proper, flawless init of the DRAM is important for the boot to happen. Imagine your un-initialized / unverified DRAM bricking DURING your boot process - it can very well be rendered unrecoverable. The SPL is then discarded once it runs.
Finally, we have the MLO or Memory LOader. This includes a basic U-boot, zImage, and the Device Tree Binary for the processor. These binary files are loaded onto the memory pre-emptively for the kernel. When the kernel init occurs, it uses these files to initialize what is essentially the configured hardware.
Stage 2 has the U-boot proper and the kernel itself.
The stage 1 ensured that all hardware and memory elements are initialized and ready. This stage, is where the bootloader (often U-Boot) and the kernel itself are loaded onto the memory of the board.
First, we have the U-Boot. It loads the kernel onto the memory, the DTB onto the memory, and holds env variables and scripts. It enables the storage and network accesses for the device, and tells the kernel where to find the rootFS.
The kernel itself, holds code that controls the CPU subsystems, the memory management unit's control ssoftware, the scheduler, init controller, device drivers, and then the specific drivers from the DTB - along with so much more.
At this point, the SBC is all initialized, memory elements holding all pieces of software, CPU fully up and running, network interfaces ready to be accessed - all that is missing, is the link between the user and these interfaces to interact. What might that be?
Stage 3 - the RootFS.
This is the link that ties the user to the world of the SBC. It holds the Directories, the app data, package managers, dev tools, userspace tools, the Window manager and the Desktop environment, and quite literally anything else that we can think of, when we imagine a linux based system.
If you are still reading this, I shall conclude today's lecture with implementations of the communication protocols that are commonly seen in Microcontrollers. A very short look at the UART devices - dev/tty, the SPI devices - dev/spidev, and the I2C devices - dev/i2c-0(or 1).
This is supposed to be an hour long lecture, but I don't think they'll have an issue if I take a bit longer than that! Finally, I shall leave them with books, and links to additional resources to explore more - should their hearts wish for.
All in all - this is about to be blast for me, and I shall do my best to spark a passion for this in my students too, so that they can learn on their own, and do some cooler things than what I could in my college days. If you have read so far - I can only thank you for listening to my ramblings.
Now back to your regularly scheduled scrolling!
Luxiom has been stuck in Apple review for weeks now lol
I sent the recording they asked for.
Set up proper delete flows.
Even formed an LLP for extra professionalism.
Still waiting😭
Anyone else living this timeline right now?
Drop your longest review wait in the replies.
xitter.cf/AapathBandhava/status/…
Happy Ayudha Puje to all Masala Dose lovers
DevDiary retweeted
From 0 to 5,200 followers in 5 months and now my X Subscriptions got approved!
I started on X on May 17. By October 17, I had crossed 5,000 followers.
Here's my journey from 0 to 5K:
Month 1: 1,700 followers
Month 2: 2,400 followers
Month 3: 3,200 followers
Month 4: 4,000 followers
Month 5: 5,200 followers
But the growth wasn't just about followers.
Here's what my analytics show:
- 15.9M+ total impressions
- 5.5M impressions in the last 4 weeks
- 2.3M impressions in the last 7 days
- 189K+ total engagements
I also got monetized and started earning payouts from X.
It didn't happen overnight. I stayed consistent, engaged with other creators, experimented with different content, and learned what worked for my audience.
And now, another milestone: I've been approved for X Subscriptions! 🎉
If you're looking to grow on X, get monetized, and build an audience of your own, you can now subscribe to my account.
I'll share my journey, lessons, and what I've learned along the way.
This is just the beginning ❣️
DevDiary retweeted
You have to retrieve data without modifying the resource. But the catch is your retrieval needs to support a long list of ids. Which HTTP method will you use in your design?
A. POST
B. GET
C. PUT
D. PATCH
I tried participating in Bolt Hackathon.
I prepared a base prompt with my intent and kept facing weird issues. Bolt either ignored instructions, or kept asing "what app shall we build"😂, probably due to context length limits
And after breaking the task down, and taking it step by step.. the result was ok ig
And my boy Claude one shotted the whole thing in 1 iteration. ONE iteration. And the vision is eerily close to what I had in mind.
Once you start using the big bois, it becomes incredibly hard to use the smaller models. Maybe Bolt's models could get me to this design, but it would require a lot of iterations
The only reason why people might use them is because they handle scaling(Idk what they do honestly) and infra which non-tech people get scared to even touch.
You have to retrieve data without modifying the resource. But the catch is your retrieval needs to support a long list of ids. Which HTTP method will you use in your design?
A. POST
B. GET
C. PUT
D. PATCH
DevDiary retweeted
The absolute wildest thing is how disconnected the majority of the world's populace is from AI
Claude Code and Codex can do whatever Lovable, Bolt, etc do but better and for cheaper even, but the only catch is you have friction(install, setup, CLI, deployment) which scares people away
Lovable, Bolt, etc charge a hefty premium, and most people dont even know the difference between models(Opus, Sol, etc). And they are ok with whatever the model gives(neon colors😆)
We on X are living in a small tech bubble.
The vast majority of the planet still use chatgpt(not Codex) for code and paste it on VSCode, or even pay for Bolt/Lovable to handle everything😂
Absolute wild times
My numbers have stabilized now, but I'm happy
I made a lot of new connections, and my old mutuals are getting my posts delivered too (to some extent)
X algo is doing its job 🙌
DevDiary retweeted
Replying to @antonships
I think the first friction is you need a desktop/laptop, somewhere where you can create persistent files
Second friction is for normies, they get scared hearing supabase, react, node.js, etc which codex and claudeuse, A LOT. Notice how Bolt use "what can we build uwu" while these guys actually grill you a little bit
Third friction is deployment, you need to handle and setup supabase/cloudflare/etc. Sure, codex already does this(computer use) but for normies this is scary("Why another website, what is database"). Or else your app will be stuck at localhost:3000
Bolt, Replit, Lovable handle the infra. They give opinionated defaults. Yes, the designs are ugly and you hit a wall, but it works and you can hand it to your friends in few minutes
The absolute wildest thing is how disconnected the majority of the world's populace is from AI
Claude Code and Codex can do whatever Lovable, Bolt, etc do but better and for cheaper even, but the only catch is you have friction(install, setup, CLI, deployment) which scares people away
Lovable, Bolt, etc charge a hefty premium, and most people dont even know the difference between models(Opus, Sol, etc). And they are ok with whatever the model gives(neon colors😆)
















