Who is Muhammad Roohullah?
Muhammad Roohullah is an aspiring AI researcher from Pakistan who is learning the math behind artificial intelligence from scratch and sharing every step in public. His full name is Muhammad Roohullah Naqvi, sometimes spelled Muhammad Rohullah Naqvi. He started his journey in September 2026 with one clear aim: to understand deep learning so well that he can build it, explain it and eventually push it forward through research.
He is not a professor or a finished expert, and he does not pretend to be. What makes his work different is that he documents the learning as it happens, including the mistakes. That makes his blog useful for anyone who wants to learn math for AI without a university course to guide them.
Here are nine things to know about him and the work he is doing.
9 things to know about Muhammad Roohullah
1. He is learning math for AI from the basics, in public
He did not start with fancy models. He started with functions, limits and continuity, then moved into derivatives and calculus. Every concept he locks in becomes a blog post written in plain language, so readers can learn alongside him.
2. His goal is to become a deep learning engineer and a researcher
Using AI tools was never enough for him. He wants to understand what happens inside a neural network, build models himself, and later write research papers of his own.
3. He follows a strict six month roadmap
His plan runs from 14 September 2026 to March 2027. It covers functions, linear algebra, single and multivariable calculus, probability, statistics and optimization, followed by modern architectures such as CNNs, RNNs and Transformers. He shares the full roadmap on this official site.
4. He studies at the library every day
From November 2026 his routine is simple: a full day at the library, roughly from 9 in the morning until late afternoon, working on both math and Python. No shortcuts, just daily hours that add up.
5. He learns math and Python side by side
Every concept he studies on paper is tested in code. He uses NumPy, SymPy, SciPy and Matplotlib to see the math working, and plans to build deep learning models from scratch in PyTorch instead of copying ready made code.
6. He decodes every symbol before he solves anything
Greek letters and dense notation scare a lot of beginners away from AI. His rule is to translate every symbol into plain words first, picture what it means, and only then work through the numbers. His posts follow the same order.
7. He turns his mistakes into lessons
He keeps an error log of every mistake and reviews it at the end of each month. One repeated mistake, mixing up the height of a function with its slope, became one of his most useful posts: ReLu is just a piecewise function . He also explains why derivatives are the language of neural networks.
8. He is building LogLens, launching in April 2027
LogLens is his own project for log pattern analysis using vector embeddings, anomaly detection and custom optimization. It is the real test of everything he is learning, built from first principles rather than by wrapping someone else's API.
9. His long term aim is research that helps people
Beyond engineering, he wants to research mathematics and deep learning, and use them to find solutions for medical diseases. For him, the math is not the finish line. It is the foundation for work that could matter to real lives.
Frequently asked questions
Who is Muhammad Roohullah?
Muhammad Roohullah is an aspiring AI researcher from Pakistan who is learning math for AI from scratch and sharing every step of his journey on his blog.
Is Muhammad Roohullah an AI researcher?
Not yet, and he says so openly. He is training to become one, starting with the math behind deep learning, and plans to research deep learning and solutions for medical diseases.
What is LogLens?
LogLens is his own project for Ai in Healthcare project, planned for launch in April 2027.
Is his name also spelled Muhammad Rohullah Naqvi?
Yes. His full name is Muhammad Roohullah Naqvi, and it is sometimes spelled Muhammad Rohullah Naqvi. All of these refer to the same person and this blog.
Follow his journey
If you want to learn math for AI the honest way, one concept at a time with every symbol explained, this blog is a good place to start. New posts arrive as each concept is learned and tested, and the roadmap is updated as the journey moves forward. on this official site you can find his code and notes are shared as he builds toward LogLens.