Writing

How Hermes Agent's memory works, after reading the memory internals of ChatGPT, Claude Code, OpenClaw, Clawdbot and smaller agent harnesses.

How Claude Code remembers things, from reading its leaked source, community write-ups and the architecture doc I built from both.

Machine Learning-এর research paper পড়তে গিয়ে প্রথমে ব্যর্থ হওয়া, তারপর ধীরে ধীরে পড়তে শেখার গল্প।

The intuition behind attention in seq2seq models: why one fixed context vector falls short, and how attention weighs every encoder state.

How encoder-decoder (seq2seq) models translate language: the encoder, the decoder, the context vector and LSTMs, with math and diagrams.

Word2Vec from scratch: why one-hot vectors fall short, dense word embeddings, the skip-gram model and its math, and a full implementation.

Build a neural network from scratch: neurons, activations, the forward pass, loss, backpropagation and gradient descent, step by step.

Build a joystick-controlled RC car with two ESP32 boards and ESP-NOW: wiring, the L298N motor driver and the code, step by step.

Backpropagation worked out by hand: a neural network's forward pass, loss and every gradient, step by step with the math.

A beginner's guide to NumPy: creating arrays and matrices, indexing, broadcasting and fast math for numerical computing in Python.

Python decorators from the ground up, starting with how names and functions work, then writing and using your own decorators.

Python generators explained with examples: how they work, why they save memory, common use cases and best practices.

Iterators in Python explained: how iteration works under the hood, and how to create and use your own iterators effectively.

A complete guide to list, dictionary and set comprehensions in Python, with examples for writing shorter, cleaner code.