Goyalayus

Blog

Machine Learning

Jul 14, 2026my notes on RL
Jun 14, 2026leanings from my first RL training run or maybe runs — my experience training 0.6b qwen model to play wordle and observing it learn to cheat and collapse
Dec 10, 2025Inference — notes for myself
Dec 9, 2025BROADCASTING — notes for myself
Dec 8, 2025MY COMPLETE NOTES ON DISTRIBUTED TRAINING — these are just notes for myself, it might feel ai written because some part of it is.
Dec 7, 2025COMPUTING GRADIENTS FOR THE SAKE OF IT — .
Nov 29, 2025CUDA Kernels Zero to One — learn basics of gpu programming and writing custom cuda kernels
Nov 16, 2025Mathematics behind Exploding and Vanishing Gradients — .
Sep 24, 2025the hitchhiker’s guide to working with Audio Data — audio data has two parts amplitude and frequency.
Aug 29, 2025Ensuring A Stable Pre-Train — there are majorly two reasons for unstable training of your llm
Aug 29, 2025You see Neural Nets Wrong — Deriving and Memorizing Forward Pass in Neural Networks
Aug 19, 2025How Multi-Model LLM's Work — To be honest it’s really simple, but please read my previous posts on Transformer Architecture and Tokenizers.
Aug 19, 2025How LLM Tokenizers Work — Understanding Byte-Pair Encoding (BPE)
Aug 19, 2025How KV Caching Works in Attention — Explained with Examples
Aug 19, 2025Tell me how you work, GPT — LLM Internals for Developers
Aug 3, 2025The Counter-Intuitive Path to Machine Learning — you do not need mathematics.
Jul 18, 2025Netflix's Big Bet — One model to rule recommendations

Systems

Oct 13, 2025Everyone is rewriting their C++ code into Rust. Why? No, It's Not Performance — .
Sep 24, 2025the hitchhiker’s guide to golang concurrency — Go has special Threads called Go-Routines.
Sep 22, 2025Thread Synchronization in C — I don’t write for absolute beginners so this post is for someone who know on a high level how threads work but does not know the low levels

Blockchain

Jul 14, 2026okay here's your Eth, but — How Lending in Decentralised level work at the level of code
Aug 19, 2025How are Stable Coins Stable — The core algorithm behind Stable Coins easily explained
Aug 18, 2025so you want some ethereum — All about how Decentralized Exchanges really work
Aug 18, 2025NFT's are Okayish — all about non-fungible tokens
Aug 18, 2025why so many coins — Fungible Tokens and ERC - 20
Aug 18, 2025Layer 2 Network Internals On Blockchain — The incomplete solution to scale Blockchain
Aug 16, 2025How Ethereum Really Works — Ethereum internals from accounts and gas to consensus and execution.
Aug 15, 2025Blockchain Internals for Developers — worlds best blog for not-so-technical folks

Startups

Aug 20, 2025So, You Want To Make A Dating App... — Our efforts for an app, which apparently never made it to the 'peepl'
Aug 19, 2025How B2C Will Play Out After LLMS — note : This post is about B2C, I have very little Experience on B2B but still will include a section at last on my views
Aug 19, 2025Paras Chopra Saved My Failed Startup — My notes from paras Chopra's Mental Models
Aug 19, 20252 Startups — Why Startup Advice sometimes feel contradictory

Life

Jul 16, 2025Why death and taxes will soon be obsolete. — .

Notes

Aug 3, 2026From CPU Hardware to Concurrency, Threads, Coroutines, and Async I/O: The Hitchhiker's Guide to Concurrency
Aug 1, 2026Learning to summarize from human feedback — my notes on this paper.
Jul 31, 2026Asynchronous RLHF, Faster and More Efficient Off-Policy RL for Language Models — my notes on this paper.
Jul 14, 2026tesla is tarding at $335.16 OR IS IT? — How are stock prices really calculated
Jul 14, 2026Reinforcement Le
Jul 14, 2026R
Jun 25, 2026Probablity
Jun 25, 2026Calculas — no more summarizable
Dec 7, 2025Block Matrice Multiplication — notes for myself
Aug 19, 2025Distribution for PMF — This blog was written before I used substack to write.