MP

My Personal Rumbling

Verified ownerAI LearnerBuilderBookish@rum4randomnessHby Hawi

A personal notebook that wanders from poems and book lists into hands-on machine-learning builds.

Hi there! This is Hawi, a person in process of becoming, AI enthusiast who is addicted to books.

MP
My Personal Rumbling
@rum4randomness· 138
Open

Hawi’s channel started as informal rumbling: short observations, tiny poems, book recommendations, and everyday thoughts. Over time, the wandering feed picked up a serious technical thread, moving through machine-learning experiments, a neural network built from scratch and now working on Hopfield Networks

Posting activity
22posts a week
396 posts in 26 weeks
26 weeks agothis week
Active since Nov 2024Longest streak 39 wk39-week streak going
Busiest time
All Hours
Peaks around 8am
Projects
2 shipped
2 tracked in posts
Language
English
90% English · 5% Amharic
Posts read
1,069
Favourite emoji 😭
  • Has posted for 39 consecutive weeks.
  • Averages 21.5 posts per week over the last 90 days.
  • 90% of text posts are in English.
  • The busiest posting hour is 08:00.

The journey.

  1. Nov 2024 – Jul 2026

    Poems, Books, and Big Questions

    The early channel was a stream of short observations, original poems, quotations, reading lists, and cultural notes. Books became a steady thread, from mystery recommendations to reflections on stories, characters, and changing perspectives.

    Sources#35#905#910#1282
  2. Sep 2026

    From XOR to CNNs

    September became a hands-on experiment lab: a neural network solved XOR from scratch with NumPy, optimizers were compared, and a grayscale CNN was completed before moving to RGB flower classification. The channel also highlighted comY and the MVP launch of AAU Exams.

    Sources#2181#2209#2255#2254
  3. Oct 2026

    Learning in Public

    The latest chapter pairs technical curiosity with questions about purpose, ambition, planning, perfectionism, and education. Hawi also makes the case for connecting school subjects rather than treating science as a set of sealed-off boxes.

    Sources#2316#2331#2330#2287

Their projects.

  • SMS Spam Classifier

    A machine learning model that detects spam text messages using TF-IDF and logistic regression.

    Launched2 posts
    Aug 2026
  • KNN for breast cancer classification

    A mini project using K-nearest neighbors to classify breast cancer cases.

    Launched1 posts
    Aug 2026

Start here.

  1. 01

    The practical ML build where the SMS Spam Classifier enters the story.

    So, as for starting this week, I created an SMS Spam Classifier project, which is a simple ML model that detects spam text messages using TF-IDF + Logistic Regr...

    Aug 18, 2026102 views
  2. 02

    A glimpse at the visual, experimental side of the KNN classification project.

    The ML Roadmap is finally live. When I started learning ML, the hardest part wasn't the math or the code. It was not knowing what to learn next. I ended up lear...

    Aug 22, 202657 views
  3. 03

    The neural-network learning thread, including solving XOR from scratch with NumPy.

    After postponing it for a while, I finally made a neural network learn XOR from scratch, using only NumPy. XOR is one of those tiny problems that teaches a rea...

    Sep 7, 2026130 views
  4. 04

    A warm example of the channel’s bookish, highly personal recommendation style.

    I finished this book last Friday . And it was awesome. a friend of mine recommended me(i decided i will trust her with my soul with her recommendations - she h...

    Mar 31, 202521 views
  5. 05

    A sharp reflection on how repeated exposure can make harmful things feel normal.

    sometimes we dont live on the standards of what's right and what's wrong, we live in a standards what we are used to long enough. the difference between somethi...

    Oct 15, 202690 views

In their own words.

we do it scared too i guess🤧
Oct 26, 2026 ↗
Attention is the rarest and purest form of generosity
May 31, 2026 ↗

Fresh from the channel.

Written from public posts, updated Oct 27, 2026. Every line links to its source.