Now that https://t.co/iUoAm9J1Ag is at a pretty stable stage, here's a view at what all I did at TrenTorch as a Founding Engineer, Maintainer & Operations Head
But first of all, WE are a very healthy team of 4 core members, each doing their part immensely well from handling engineering, infrastructure & maintenance to working on growth & outreach
As is evident from my posts and project work, I'm into ripping systems apart & building infra around those.
So, at TrenTorch:
1. When we started off with the project, we derived the CLI version from TinyTorch (as already stated everywhere). I went through the whole system, simplified it, and built a gorgeous interactive TUI around it for better accessibility
2. I added multi-format conversion support with runtime platform detection, allowing users to run the modules on standard notebook platforms and in sandbox environments
3. To make this even more "open" for all, we made https://t.co/iUoAm9J1Ag from scratch. I built the web-based IDE that's up in there
4. We provide our users with an extensive curriculum. I'm responsible for it to be damn clean, succinct & purposeful (no redundency) while providing a rather easy-to-digest progression through the tracks
5. We regularly work on the UI to give our users the best experience. I test this rigorously, ensuring nothing breaks while doing so, and that it matches our standards
6. Following the recently added support for SQL & python visualization libraries, ensuring it all runs well while still rendering beautiful interactive widgets was my responsibility
7. Recently, I've been more so involved on the reliability side of things to minimize the number of mishaps on TrenTorch
On a microscopic scale, there are a lot of things that are not really summarizable here
Will have an article dedicated to that rolling out soon
All-in-all, I handle this loop of maintenance, feature pushes & bug fixes, GitHub issues & actions, and site reliability. Building subsystems and pulling in refinements made by other team members, all while ensuring the whole system remains stable as it evolves
For context:
β’ We have our Chief Commanding Officer, @Rocky_T07 who is also THE Founder, orchestrates the entire team to make sure we all are aligned, acing our tasks and killing it on daily-basis while he continues to architect better & better infrastructure
β’ @aadityansha_06 is our Co-Founder & Product Head ensuring TrenTorch is known & used at scale, while also bringing in relevant support to keep all this running for free, for everyone
β’ @maanas_tyagi is our Core Engineer who gets the initial part of stuff gathered so I can work through it. Setting the first stage so the team doesn't have to wander around from the very start
Well, we just made this damn dense, yet gorgeous & digestible as always! β¨
We've got:
- 11 systematically curated tracks
- support for python libraries external to pyodide
- support for SQL & data visualization
- a total of 1287 questions
- 250+ questions dedicated to company interviews
- an airy UI with lots of character
done this far
What that means for you, is:
1. You get a very clean, laid out structure to study from. Choose a specialization track, skim through or go deep, your call!
2. You can now build plots, create tables, manipulate those and see everything happening live on screen with little to no latency
3. You can prepare yourself, not just for the systems, but now also for DS/ML/AI job interviews
TrenTorch is getting better every single day with increasingly more number of users β¨
That motivates us to make this even better, ofc! βοΈ
Nah, we keep doing such things to make sure our users @TrenTorch never have to experience any latency at all, lol! π
On https://t.co/iUoAm9JzpO, you can revisit concepts & progress towards hard questions in your selected track OR get deep into blogs & research papers
Well, if ydk what all this is, ig it's time to step-up before your math, data & systems fundamentals start "rusting" too π
Engineering is a lot about cleverly faking the process
The mechanism underneath isn't necessarily the one people intuitively imagine
And it doesn't have to..!
What matters at the end of the day is obtaining the right result. At this point, preserving the properties we care about at that abstraction level is all that's needed
We see this everywhere:
β’ Digital logic operating on switches that themselves could be based off anything from physical holes on a card to eβ» holes in a silicon molecule
β’ A randomized algorithm deliberately introduces randomness into a process whose desired statistical behavior is known
β’ An equation is a succinct generative description whose solution happens to mathematically exhibit the behaviour we are interested in
β’ Cryptography constructs such computational problems that recovering information is infeasible under specified assumptions
β’ Control Engineering, Semiconductors, Boolean Algebra, Numerical Methods, Simulations, Neural Networks, Programmer chips,...
Making a thing impersonate another without needing to become it β that's the core idea
The mechanism then becomes yet another tool
I'm very happy to announce that TrenTorch has been selected for Vercel's open source program (Summer-2026).
Happy to grow under Vercel's guidance and support !!
@maanas_tyagi@OpenAI Yeah, there's no daily or weekly limit π«
Surely made for some last-min late builds π so you have your whole month stacked together
Sustainability engineering is, at its core, the design of near-closed resource loops
Not merely making each component more efficient, but designing the system so that the output of one subsystem becomes an input to another, keeping energy & material flowing through useful pathways for as long as possible
The goal isn't to eliminate losses
That's just impossible
It's to minimize irreversible losses by giving as much of the system's output as possible somewhere useful to go
This applies far beyond conventional green engineering
β’ Nature does it
β’ Industrial ecosystems can do it
β’ Cities & Economies
β’ Computing infrastructure
β’ Spacecrafts
β’ Biological systems
And countless other complex systems can be understood through the same lens
The deeper systems-engineering principle is:
Don't optimize components in isolation
Optimize the flow through the system
An output becomes waste only when the system has no utilizable pathway for it
We are helping people build capabilities in frontier tech @ https://t.co/iUoAm9JzpO
Democratizing tech education, w/o any geopolitical bounds π
You'll learn systems, algorithms & infra (learning algorithms, high-performance computing & eventually hardware-centric systems) through two gorgeous interfaces: TUI & Web, whichever is your thing
We brush up the theory with micro-lessons beside the IDE, so you're learning the concepts while actually building with them
This is us trying to build the machinery for our philosophy:
Knowledge β Capability β Work β Opportunity
Don't chase Lakshmi (money)
Flirt with her enough to keep her in the "interest" loop, and go on regular dates with Saraswati
The capabilities you build by interacting with the second help you get the first one πΆβπ«οΈ