15 / 3 / 2026

---

TASK 2 : API

An API (Application Programming Interface) is a set of rules and protocols that allows different software applications to communicate with each other. I used the OpenWeather API — a free weather data service — to build a live weather app in HTML, CSS and JavaScript that fetches real-time conditions like temperature, humidity, and wind speed for any city and displays them on the UI.

TASK 3 : Working with Github

GitHub hosts Git repositories and gives you a central place to store code, track changes, and work with others. Actions automates things like testing and deployment, Issues keep track of bugs and tasks, and Pull Requests are how you propose changes before they get merged.

• Visited UVCE-Marvel/git-task and read through the README.
• Forked and cloned the repository to make changes locally.
• Created a new branch to keep the changes isolated from main.
• Opened a Pull Request, reviewed the diff, and submitted it.
• Looked through existing Issues to understand how they work. link

TASK 4 : Getting familiar with command line

Ubuntu is a Linux-based OS that is open-source and free. It has a GUI but the terminal is where it gets genuinely useful — file management, automation, and system tasks are all faster through the command line.

Went through the beginner's guide first, then carried out the following on the terminal:

• Created a folder called test using mkdir
• Moved into it with cd test
• Created a blank file without a text editor using touch
• Listed files with ls
• Created 2600 folders named with a letter-number pattern like A90 using a loop
• Added text to two .txt files using echo
• Concatenated and printed both files using cat

mkdir test
cd test
touch file1.txt file2.txt
ls
mkdir A{1..2600}
echo "this is file one" > file1.txt
echo "this is file two" > file2.txt
cat file1.txt file2.txt

TASK 5 : Build your own Brain (Linear Regression)

What is Linear Regression?

You have a bunch of scattered data points and no single line can pass through all of them. Linear Regression finds the best fit line — one that gets as close as possible to all of them. The gap between where a point actually is and where the line predicts it to be is the error, and the whole point of regression is to shrink that error. It is the simplest prediction algorithm in ML.

---

The Math Behind It

A straight line is y = ωx + b, where ω is the slope (weights in ML) and b is the intercept (bias). The job is to find the ω and b that keep prediction errors as small as possible across all the data points.

---

Loss Function

The loss function tells you how wrong the model is:

• MSE (Mean Squared Error) — squares the errors before averaging
• MAE (Mean Absolute Error) — uses absolute values instead

MSE is the go-to because it differentiates cleanly, which makes the math behind training much easier to work with. Linear regression

---

TASK 6 : The Matrix Puzzle (Getting familiar with numpy)

NumPy handles numerical computing in Python — multi-dimensional arrays, matrix operations, the works. Matplotlib takes care of visualization. Together they are the standard starting point for any data work in Python.

Watched the tutorials first, then worked through the puzzle in Google Colab:

1. Installed and imported NumPy and Matplotlib
2. Loaded the .npy scrambled array
3. Checked the shape and reshaped it into a square matrix
4. Rotated it 90 degrees to fix the orientation
5. Used matplotlib.pyplot.imshow() to reveal the image

Click here to view the notebook

TASK 8 : Markdown

Markdown is a lightweight markup language that formats plain text and converts it to valid HTML, consistently across any device. I learned to structure a document with headings, body text, and embedded images, videos, and GIFs, then wrote a report on F1 to put it into practice. Article

TASK 9 : Tinkercad

Tinkercad is a free browser-based tool by Autodesk for simulating circuits and designing 3D models, no hardware needed.

Built a circuit with an ultrasonic sensor and Arduino that calculates distance by measuring how long sound waves take to bounce back, with results shown on the serial monitor. Then added a servo motor to turn it into a radar system that sweeps the sensor across angles for wider coverage. Both circuits simulated and working as expected. Click here to view the github repo

TASK 10 : Speed Control of DC Motor

Two things control a DC motor: PWM for speed, by switching power on and off rapidly to adjust average voltage, and an H-Bridge for direction, by reversing current flow. The L298N sits in between, taking the Arduino's weak signals and amplifying them enough to actually drive the motor.

Understanding motor control

• PWM (pulse width modulation ) — controls speed
• H-bridge — controls direction

What exactly is H-bridge?

Four transistors acting as switches, two closing at a time to decide which way the motor spins. Click the image to view the simulation

TASK 11 : LED Toggle Using ESP32

The ESP32 is a cheap microcontroller with built-in Wi-Fi and Bluetooth, programmable through the Arduino IDE. The task was to run a web server on the ESP32 itself and control an LED from a browser on the same network.

Set up the Arduino IDE with the ESP32 board package, wired the LED to a GPIO pin with a resistor, and uploaded a sketch that connects to Wi-Fi and starts a web server. Opening the IP address in a browser loads a toggle button — clicking it sends an HTTP GET request back to the ESP32, which flips the GPIO pin and serves an updated page. LED toggling wirelessly, task done.

Click the image to view the simulation

TASK 12 : Solder Prerequisites

Soldering joins components by melting solder around connection points, making a permanent bond that can be undone with desoldering when needed. Got introduced to the equipment — iron, solder wire, flux, wick, and stand — and practiced soldering and desoldering on a perf board under coordinator supervision.

TASK 14: Karnaugh Maps and Deriving the logic circuit

K-Maps cut down complex boolean expressions by grouping terms into pairs, quads, or octets to get the simplest possible logic circuit.

Built a burglar alarm with two inputs — door status and key status. The K-map showed the output is 1 only when D = 1 and K = 0, giving D·K̅. The alarm goes off when the door is open and the key is not being used. Click the image to view the simulation

TASK 16 : Datasheet report writing

Went through the L293D datasheet to understand what is actually inside the IC — transistors, logic gates, and protection circuits. The H-bridge handles direction by flipping current flow, and PWM handles speed by switching power on and off. Together they give the chip full control over two motors. Report

TASK 17 : Introduction to VR

Virtual Reality puts you inside a simulated 3D environment, making you feel like you are somewhere you are not. Report on VR

TASK 18 : Sad Servers "Like LeetCode for Linux"

The Command Line Murders is a detective challenge — find the culprit buried in system files and logs using only Linux commands.

Read through the Linux commands article first, then launched the browser terminal and got to work. Used ls and cd to navigate, cat to read clue files, and grep to filter through logs for anything useful. Tracked down the evidence, identified the perpetrator, and echoed the name as the answer to make the server happy again :) .

Task 20 : Notebook Ninja – Getting Started with Jupyter

Jupyter lets you write code, notes, and charts all in one file. Put together a notebook with Markdown for structure — intro, objectives, summary — and Python for a line graph showing monthly sales. My Jupyter Notebook

Task 21 : Watch and Reflect - Intro to Machine Learning

The main thing I took away from both videos: the model is only as good as the data going into it — Garbage In, Garbage Out. StatQuest made it clear that fitting training data perfectly does not make a good model — predictions on new data are what actually count. AltexSoft showed the other side of that, covering how data needs to be cleaned and structured through labelling, reduction, and normalization before training even starts, and why splitting into train and test sets matters for honest evaluation. Read further in the article