Ideas thatrefuse to sit still.
Essays on AI fairness, machine learning, geopolitics, software craft, and the quiet politics of how we build things. By Mukul Namagiri.
The Writing
Math for Machine learning
Machine learning from a formal view
Machine learning is the study of algorithms that improve automatically through experience.
I tried to put together Machine learning from statistical learning theory and introduced Ml from a rigorous mathematical stand point
Read →"Understanding Generative and Discriminative models"
Origin of reinforcement
A condensed, symbol‑dense reference covering the mathematical foundations of reinforcement learning
Read → Probability and statisticsIntroduction to statistics for Machine learning
A structured mathematical reference covering probability spaces, random variables, conditional probability and expectation, measure-theoretic foundations, estimation theory, and the statistical tools
Read → Linear algebraIntroduction to Linear algebra for Machine learning
A structured mathematical reference covering vector spaces, linear transformations, matrices, determinants, eigenvalues, the SVD, and the optimization theory
Read → CalculusIntroduction to Calculus and optimization theory for Machine learning
A structured mathematical reference covering limits, derivatives, integrals, sequences, series, multivariable calculus, differentials, matrix calculus, and Taylor's theorem — extended with machine learning essentials:
Read essay →The Writing
Essays, Opinions & Deep Dives
A collection of long-form thinking across AI fairness, machine learning theory, responsible scaling, software archaeology, and the geopolitics of intelligence.
A detailed breakdown of transformer architecture
Understanding the internal workings of the llms is important and a rapidly evolving field as a precursor we did a breakdown of gpt 2/3 to understand how these systems actually work
Read essay →"In this section we analysed all aspects of attention mechanism"
A Healthy AI is More Accurate
On normalization — the quiet discipline that keeps neural networks from growing unchecked. Drawn from linear algebra and biology, this is about why controlled scaling matters more than raw power.
Read essay → Machine LearningWell, Well, Well: It Turns Out Machines Are Also Bigots
Microsoft's Fairlearn enters the ring, armed with charts, good intentions, and a frank acknowledgment that it cannot fix everything — much like us. A sharp, satirical take on how human bias gets laundered into AI datasets with admirable efficiency.
Read essay → Software CraftThe Story of Broken Structures
On the quiet collapse of codebases, the illusion of order, and why a cheat sheet might just save us all. A meditation for the weary debugger who arrived from pure science and found beautiful chaos.
Read essay → AI GovernanceHow to Grow AI? — Scaling Carefully
AI is an emergent phenomenon, not a developed technology. On responsible scaling policies, the nuclear analogy, and what happens when something this capable is released before we fully understand it.
Read essay → Geopolitics & AIAI Diplomacy — From Coins to Cognition
On trust, trade, and the strange new currencies of model weights being minted at the edge of human intelligence. From Mesopotamian grain trades to AI geopolitics — the question beneath every transaction.
Read essay → Agentic AIAgents: The Age of Action
On agentic AI systems, introspection, and what it means for machines to move from thinking to doing. A personal essay on slow deliberation in a world rewarding quick reflexes.
Read essay → AI ThinkingThinking fast and slow looking inside ai thought process
A fine summer in Dartmouth and the events leading upto the dicovery of chain of thought and the procedure to safe guard when chain of thought is exposed
Read essay → A Thought ExperimentWho can train a perfect AI ?
We reimagine a situation in which Winston from 1984 is given the task to design an AI
Read essay → AI monologueSinister and Saint AI
Are we gonna have mechanical gods or Helpful Servants only time will tell Hear the Ramblings of AIs
Read essay → Existencal RisksNuclear, Cyber and Bio weapons
I focused on nuclear, biological, and cyber domains, and documented a series of incidents that illuminate something crucial. Not just the mechanics of catastrophe, but the deeper patterns of human nature driving them.
Read essay → Introduction to AI safetywhat, why and How
A simple introductory guide on ai safety uysing 13 different topics
Read essay →Projects
Things I've Built
Live applications exploring the intersection of information, fairness, and intelligence.
Factapp
A live web application — home to essays, experiments, and ideas on AI, agency, and the question of what machines should be allowed to decide.
Visit app →Form Project
A hosted form application built on PythonAnywhere — part of an ongoing effort to make data collection thoughtful and accessible.
Open form →Factity — The Unfair Observer
A long-form publication on AI fairness, bias, and the societal stakes of algorithmic decision-making. Essays that implicate the reader.
Read publication →Portfolio
Work & Profile
A curated view of projects, skills, and background — built for those who want the full picture.
Mukul Namagiri — Portfolio
Research, writing, and technical work across machine learning, AI fairness, and software. The full professional profile.
View Portfolio →Support
Consider Supporting the Work
Good writing takes time. If any of this thinking has been useful, interesting, or even just made you pause — a small contribution keeps it going.
Every contribution is deeply appreciated
Click the button or scan the QR code — whichever is easier.
.png)
Scan to support