✦Welcome to my blog
Hi, I'm Tusher
This platform documents what I'm learning, what I'm building, and how my technical direction evolves over time.
✦Featured writing
Recent writing that reflects how the work is evolving
A mix of learning notes, project thinking, and system-building reflections.
What I wish I knew before writing my first Python ML pipeline
Practical lessons from the mistakes nobody warns you about — data leakage, silent bugs in preprocessing, and why your test accuracy is probably lying to you.
✦Across the site
Recent signals from writing, study, and curation
The homepage now reflects the full shape of the platform, not just the blog archive.
Featured writing
Recent writing that reflects how the work is evolving
A mix of learning notes, project thinking, and system-building reflections.
Python • May 18, 2026
What I wish I knew before writing my first Python ML pipeline
Practical lessons from the mistakes nobody warns you about — data leakage, silent bugs in preprocessing, and why your test accuracy is probably lying to you.
AI / ML • May 18, 2026
The difference between reading about gradient descent and actually understanding it
Gradient descent shows up in every ML explanation. Most of those explanations tell you what it is without helping you feel why it works. This post tries to close that gap.
Academic and research
Research notes, experiments, and academic continuity
A space for paper-reading, coursework reflections, research interests, and later thesis work.
paper note • Mar 2, 2026
Paper-reading workflow for ML and LLM topics
A lightweight process for reading papers with better retention and clearer downstream experiments.
research note • Draft
Notes on "Attention Is All You Need" — reading the original transformer paper properly
Reading notes on Vaswani et al. (2017) — the paper that introduced the transformer. Covering the key architectural decisions, what the notation actually means, and the questions this paper leaves open.
Recommendations
Resources worth recommending because they genuinely help
Tools, books, and learning assets that support real progress instead of hype.
✦Recent updates
Fresh notes and visible progress
Recent updates remain automatically driven by published posts so the homepage stays honest and self refreshing.
The difference between reading about gradient descent and actually understanding it
Gradient descent shows up in every ML explanation. Most of those explanations tell you what it is without helping you feel why it works. This post tries to close that gap.
Building My AI Learning System (Instead of Chasing Tutorials)
A practical approach to building a structured AI learning system instead of endlessly consuming random tutorials.
✦Trending Topics
Ideas on the move
Derived organically from what I'm writing about.
✦Connect
If you care about AI, ML, systems, or serious learning, let us talk.
I am building this platform as a public record of growth. If there is a research idea, project, or conversation worth having, reach out.
The newsletter is not live yet, so this site points you to real ways to stay in touch instead of collecting emails without a working pipeline.