Abhishek Parashar abhishek-parashar
~ / profile.sh - bash
$ whoami

> Abhishek Parashar

// engineer · researcher · pattern_hunter()

Software engineer working on machine learning and AI research, and writing about it in language that doesn't require a PhD to read.

$ ls /actions
cat about.md

A quick introduction.

I'm a software engineer with a deep passion for coding and AI research. I also have a habit of finding patterns in everything, in data, in history, in human behavior, and yes, in the way coffee tastes slightly different every morning.

Much of my time outside work is spent reading, philosophy, history, and polity. I'm drawn to the big questions: how civilizations rise and fall, how ideas shape societies, and why we think the way we do. These interests naturally led me to history at large, old texts, monuments, artifacts, and the centuries of evolution that shaped the rich, diverse culture of our country. Understanding that journey is something I take genuine pride in.

I also play guitar and sing along, I try my level best to make it sound like music. I also dance my heart out on Bollywood chartbusters.

Like most Indians, cricket is sacred to me. I follow every match and enjoy breaking down the strategy behind every session. If over-analyzing cricket were a discipline, I'd probably publish papers on it.

Over the past few years, I've immersed myself in machine learning and deep learning, from the models themselves to the computational mathematics that powers them. The deeper I go, the clearer it becomes that AI is as much a mystery as it is a science. That realization is partly what drives me to write, to cut through the jargon and make the complex foundations of machine learning truly accessible. Someone has to translate "stochastic gradient descent" into human, and I've volunteered.

I write mostly about machine learning and data science, and occasionally about history and polity. You'll find my blogs linked on this site.

When I'm not writing code or writing about it, I'm probably staring at the night sky, quietly humbled by how much we've yet to understand, and mildly frustrated that the universe refuses to show its source code.

Find me on GitHub, Twitter, and LinkedIn, I'm always open to collaboration, conversation, and good ideas.

ls ~/writing | head -3

Recent writing.

prompting-as-storytelling.md
Prompting as Storytelling
Understanding how language models work and why the best prompts share the same principles as good storytelling, from vector embeddings to attention mechanisms, audience specification to few-shot learning.
cat post.md →
kashmir-part-1.md
The Story of Kashmir: Part 1
From its ancient origins and the legend of Rishi Kashyap, through independence, the Baramulla massacre, the Instrument of Accession, and UN Resolution 47, the untold story of Kashmir told through official documents and facts.
cat post.md →
bring-back-the-colours.md
Bring Back The Colours
Colourizing old black and white images and videos using Deep Learning, a walkthrough of autoencoders, the colourization pipeline, and results using a fast.ai pretrained model on Google Colab.
cat post.md →
$ ls ~/writing, see all posts →