Selected projects in Deep Learning, Machine Learning

---

Building Neural Network From Scratch

The project is about creating a neural network from scratch, without using external libraries. The goal is to explain the loss.backward() function in detail, by solving the equation “2* ? = 4” and finding the missing value (?). To improve the network’s performance, we’ll be using learning rate decay and backpropagation techniques.

Skills Used -> Python, Pytorch, Linear algebra, Calculus

Source_Code

---

Created a SkimLit Model With Tensorflow

I implemented SkimLit PubMed AI model using transformer architecture and achieved an accuracy of 84% on all accuracy metrics precision, recall, f1_score, accuracy. The Whole Model and The whole output is in my github repo.

Skills Used -> Python, Tensorflow, Keras, Pandas, Matplotlib, Flask, html, css

Source_Code

---

classifying satellite images

I developed a satellite image classification model that can identify four different types of terrain - forest, desert, water, and cloudy - with an impressive accuracy rate of 98.7%.And i’m not gonna lie This is the final result of my model in confusion matrix.

Skills Used -> Python, Tensorflow, Kaggle Dataset, Pandas, Matplotlib

Source_Code

---

Scraperella: To collect information from the web for research purposes

I created a web scraping bot that stalks people and organizations, gathering their information and saving it in an HTML format. It also has a hobby of collecting images and saving them in a separate folder for deep learning purposes. The bot’s UI is in HTML, and it’s powered by a Python Flask API. Don’t worry, the bot won’t get a restraining order (yet)

Skills Used -> Python, Selenium, Beautiful Soup, Pandas, HTML, CSS, Flask

Source_Code

---

Ohh! I’m also a blog writer✍

justdoitanything

Skills Used -> Creativity, Understanding of ML and DL concept fundamentals, Communication Skill, Domain Knowledge, Critical Thinking

---