Learning Resources

Here are some learning resources that I find valuable:

My neurons basically dream/repeat like a stochastic parrot with content from these channels..few Science Educators i love and support! will recommend for first principles thinking. Shoutout for making complex math intuition easy!

• StatQuest with Josh Starmer
• 3Blue1Brown
• These two above: alone will basically spoon-feed you the intuition for Calculus (slopes aka derivatives and areas aka integration). Toss in some Linear Algebra for matrix wrangling and equations, and boom—you're also on your way to understanding that mind-bending higher dimensional latent space stuff!
• Great intuition on higher dimensional latent space: Visualizing high-dimensional data, as humans are restricted by duality 2D(max 3D (2D+ depth maps))
• Artem Kirsanov The Quality 🤌🏻 , why internet is amazing!
• VeritasiumDerek is amazing at turning convoluted sciencey-wordy narratives into a linear epic storylines for the layman to understand without loosing the gist
• PBS Space Time
• The Science Asylum (Nick Lucid) It's okay to be little crazy!
• ElectroBOOM Doesn't need a intro :)
• Kurzgesagt – In a Nutshell
• Vsauce
• LLM 101 Playlist by Andrej Karpathy
• bycloudAI
• Fireship
• ByteByteGo
• exurb1a
• Steve Brunton (Legendary Eigensteve - Teaches Math to Engineers) Control Theory meets ML
• Anton Petrov
• How To Be Successful by Sam Altman - insightful guide on achieving outlier success
• Arvin Ash - Great channel with summaries on complex physics/math concepts.
• Welch Labs - Excellent visual/narrative explanations of complex math concepts, particularly the series on imaginary numbers are real, NNetworks and origins of shape rotating euler's formula.
• laid back information consumption:
• Lex/JRE/Dwarkesh podcasts - few great guests(Naval Ravikant, chris lattner, jeff dean, etc.)/conversations on how to calm the mind (ackchyually! reason is: to steal the "chain of thoughts" from all the amazing humans out there/who came before me and the people who are still alive)
• Closer To Truth - Fascinating discussions with leading scientists and philosophers exploring fundamental questions about consciousness, the cosmos, and existence itself. (Makes you Realize "-Action- produces more information - You can just do things" is better philosophy than just thinking and being a P-zombie)
• Theories of Everything with Curt Jaimungal - A little bit out there into Woo Science but great guests and conversations (gotta believe in the pareto distribution of information :) )

Compute Focus:

• The Secret Lives of Data: Raft Consensus Algo - Excellent visual guide explaining how distributed systems achieve consensus/agreement through leader election, log replication, and safety guarantees. Great for understanding both computer distributed systems and real-world election/voting systems through the lens of timeouts, race conditions, and synchronization.
• Linux internals TechTalk by Ken Guyton at google dallas office i think - A 6-video playlist from 2008 covering kernel 2.x. While dated, it provides valuable insights into low-level primitives (which dint change much even today with 5.x) and the thought processes of early developers in meeting workload demands of that era.
• Viacheslav Biriukov's Blog - An excellent summary for SREs (who are lazy to go through linux kernel code and need a kickstart) or dev dealing with terminals/console/shell. The blog provides amazing summaries on shell-related internals such as file descriptors, pipes, terminals, user sessions, and process groups. It also includes good summaries on the Linux Page Cache and DNS resolvers which is crucial for understanding system performance.
• Software 2.0 by Andrej Karpathy - Insightful 2017 article about how neural networks/gradient descent is becoming better "programmer" than humans for certain(or 90% of program space?) tasks. Great perspective on the future of software development/design and hardware accelerators.
• Inference Parallelism in LLMs - Comprehensive overview of different parallelization strategies (Data, Tensor, Pipeline, Expert) for deploying large language models across multiple GPUs. Great resource for understanding the tradeoffs between latency, throughput and hardware configurations when scaling AI inference.
• GPU Performance Metrics (GPU Utilization is Misleading) - Comprehensive analysis of why traditional GPU utilization numbers don't tell the full story. Explores key factors like memory bandwidth bottlenecks, compute vs memory bound operations, and the hidden costs of kernel launches that impact real ML workload performance.
• Mechanical Sympathy Blog - Excellent blog exploring the harmony between hardware and software. Provides deep insights into how understanding the underlying hardware architecture can lead to better software design and performance optimization.

BioHacking Focus:

• Biohacking Lite by Andrej Karpathy - Decent 101 exploration into basic molecules/cells that help our gears run.
• Tracking calories is the important part (bomb calorimeter is the living proof that weight and calorie burn-rate matter ~ Calories ~= joules ~= energy ~= mass ~= weight ~= fat ? Do the math! Can't beat the quantum foam ..only way is tracking calories)
• Bryan Johnson's Blueprint Protocol - Even if the protocol gives some American psycho sigma bale vibes lol..there's few valid stuff to learn on some of the optimal thresholds (meaurements) of the most often-ly done clinical trials

Add LLMs like gpt4 speech to text whisper api.. should be able to learn basically anything by mostly just speaking without running into Carpal tunnel syndrome or needing to type

Above resources have - Decent Noise to Signal ratio - Good place to start Compressing the Data/Patterns for the ~100 Trillion Parameter model that is our brain 🙂🧠_🤖🦾

if some link is 404 , try using waybackmachine archives

Back to Timeline