Medium

Nikhil Anand

11.2K followers

Pinned

In

AI Advances

by

Nikhil Anand

·

Mar 6

This one math trick could make LLMs remember 100x more.

I thought I understood LLM memory until I learnt this.

This one math trick could make LLMs remember 100x more.

Pinned

In

AI Advances

by

Nikhil Anand

·

Sep 6, 2025

Why AI loves eigenvectors (and you should too)

Compression, clarity, and the math that makes it possible

Why AI loves eigenvectors (and you should too)

Pinned

In

AI Advances

by

Nikhil Anand

·

Mar 19, 2025

How to make your own AI software engineer (like Devin)

…using DeepSeek’s RL blueprint for everything AI.

How to make your own AI software engineer (like Devin)

Pinned

In

AI Advances

by

Nikhil Anand

·

Mar 14, 2025

After DeepSeek-R1, a goldmine of research is waiting to be discovered.

DeepSeek’s RL ideas are more game-changing than you think.

After DeepSeek-R1, a goldmine of research is waiting to be discovered.

Pinned

In

AI Advances

by

Nikhil Anand

·

Dec 2, 2024

My LLM’s outputs got 1000% better with this simple trick.

I wish I had known this trick sooner.

My LLM’s outputs got 1000% better with this simple trick.

In

AI Advances

by

Nikhil Anand

·

May 30

I spent months building an AI evaluator. Then I realised I had no idea if it worked.

Who evaluates the evaluator?

I spent months building an AI evaluator. Then I realised I had no idea if it worked.

In

AI Advances

by

Nikhil Anand

·

May 2

The math behind choosing dinner (and training AI)

From your dinner plans to billion-parameter AI models

The math behind choosing dinner (and training AI)

In

AI Advances

by

Nikhil Anand

·

Apr 11

This is why your LLM outputs keep breaking.

And no, better prompting won’t fix it.

This is why your LLM outputs keep breaking.

In

AI Advances

by

Nikhil Anand

·

Mar 21

Backpropagation is simpler than you think (once you see this)

What’s actually being computed inside a neural network (step by step)

Backpropagation is simpler than you think (once you see this)

In

AI Advances

by

Nikhil Anand

·

Feb 21

This embarrassingly simple idea explains all of AI.

How a simple math idea uncovers the real truth about AI.

This embarrassingly simple idea explains all of AI.

Nikhil Anand

Nikhil Anand

11.2K followers

AI Researcher at Adobe Research | Data Science+Bioengineering at IIT Madras | Weekly newsletter: https://makingaieasy.substack.com/

Following

Help

Status

About

Careers

Press

Blog

Store

Privacy

Rules

Terms

Text to speech