Logs '25
Jul 7 ‘25. Some links on ML and CS concepts, well illustrated. Amit’s writeup on REGEX and Self-supervised learning. Kudos on the blog design, really feels like a good technical blog. Aman.ai’s primers on KNN’s, Clustering, and Speech Processing and Shuzhan Fan’s post on ResNet’s and SVM’s.
A special note to self: Skills are what help you bring your ideas to life - no matter what (supposedly) brilliant idea you have - if you ain’t got the skills, it may not see the light of the day.
Jul 5 ‘25. Some links to follow and some mistakes that made me dive deep into what are iterators. Link1 - Biyani, Link2 - PG’s tour of Python Internals, Link3 - Iterators in C++, Link4 - Py Iterators in C, Link5 Iterators in C++ and Link6 - History of Iterators. To be continued.
There are two things I realized today - when learning, you need to follow your intuitions to the end, to figure out if they led you down the right path or not AND - you need to learn how to get your intuitions / ideas into the real world. Figure a way to express them, and test them, that is part of getting good at your craft.
Jul 3 ‘25. Found out about some of the computer vision & ML work netflix has put in to analyze edit videos to produce higher quality content and search on their platform. High-level explanation behind Match-Cut is quite interesting. Especially how they went from the segmentation x IOU technique for frame matching to Optical flow x cosine similarity for action - matching. I really love how fluidly the scene is transitioned, with instant understanding of the content. I realised something else, in terms of thinking about how you would use the varied things you’d learn in a course - say concepts like cosine-similarty and IoU, what prbably helps is looking closely into tasks & tools. Tasks are the part of the problem you are trying to solve, say similarity search between two frames OR similarity search between actions of two frames and tools are what aid you to do this - IoU for finding similarity between two different areas; which is suitable if you are comparing masks from segmentation VS cosine similarity to identify similarity between two vectors; suitable for the vector field output provided by optical flow.
Jul 1 ‘25. Turned 27, moved to Munich - ready to be poor ;). Mostly, open questions about wealth, money and paper currency is on my mind. Watched another video on the paper money, still lot’s to understand, but this one is more on the role of Central Government banks (in context of UK’s Bank, but obviously applicable to India & The US) - 97% Owned. Man, if you can’t trust “numbers in the bank”, bond’s and stocks (I hope you’ve read Value of Nothing), then what is left. You can’t buy commodities and just store them somewhere safe (Or can you ?).
The next best thing is to roll up your sleeves, and create wealth. This thing is important for literal survival of everyone. I used to think that wealth would mean “assets being held”, or “large sums of money” or vaguely “that big company / startup / business”. Wealth is, as David would say - wealth is a set of physical transformations that we can affect.
A knowledge of and ability to control your means of production of food, water, and electricity for example, is wealth. An economy essentially is exchange of goods and services such as above (that go way beyond the basics). It’s also, an attempt at improving life by solving problems, winning over nature - that’s what market’s respected.
The finance sector was born and it destroyed the very map used to keep track of this exchange. It was a way to keep in check, problems, that would cause excessive spending on defence, on waging wars, on jacking up asset prices to sell to the sucker that held it till the end, and it’s now paper floating around.
Well, Wealth, Urban Mobility design, building inter-generational stability and wealth was on my mind. As an aside theme. Also see - WTF Happened in 1971. I mean just look at this chart!
Also found another short story by the Ted Kaczynski to be quite interesting - Ship of Fools.
Ultimately, grateful for the opportunities so far and cautious of the opportunities to win over nature. And get insanely good at things.
Jun 22 ‘25. Should be studying for an upcoming exam, but went into a rabbit hole on how Fragile money really is. Some interesting links, like this cartoon on How Money Works was the first trigger. If you are looking for serious adult explanations look at the series Hidden Secrets of Money.
Jun 21 ‘25. In personal news, I’m moving to Munich, to work as a student and to work on my Master’s Thesis with Nexustec.eu, more on that later.
As part of moving, I was curious how cycle friendly is the city, the distance from where I live to the office is about 10 kms via cycling. Almost 40 mins of travel time one way. Found some links on Cycling around in Munich city - which is overhauling it’s cycling infrastructure! Munich - touring through the city - 1 and Munich - touring through the city - 2 show a good deal of what is the city currently like for cycling and what are their plans in the future.
Re-read the written work on the value of nothing, observations match with what I’m currently drafting as an essay. A strage excitement in writing the same.
The source was from one of George Hotz’s video, the longer rant on economics, american empire, china, bitcoin etc. is quite good.
Jun 20 ‘25. Started reading Maitreyan’s book Manushyarariyan - മനുഷ്യരറിയാൻ. Quite a few interesting ideas and avenues to explore.
Recently got hooked to watching the life of two canadian’s in a Japanese country side, if not for supposed peace I experience seeing the videos, it was definitely worth it for the aesthetic value of each frame shot. Kudos to that!