With the rise of edge AI and embedded intelligence, safety-critical applications like driver drowsiness detection can now be implemented on resource-constrained devices, reducing latency, power consumption, and cost compared to cloud-based solutions.

This is just the software stage for now — but the project is designed to be deployed on real hardware, making it a true end-to-end embedded system project.

I designed this tool as a final project in "Stanford University-Code in Place-2024" Python Campaign.

This tool is designed to simplify the process for electrical and electronic enthusiasts when working with circuits containing resistors. Choosing the right resistor for a circuit can be challenging, despite calculating the required values. Traditionally, we rely on the Universal Resistor Color Code Chart or a Multimeter to measure resistance. With this tool, you no longer need to memorize resistor color codes or use a Multimeter for initial usage and resistor identification. It streamlines the process, allowing you to identify resistors effortlessly. For precise circuit requirements, measuring resistance with a Multimeter is still recommended. However, if you have a box of unidentified resistors, my "Universal Resistor Color Code Identifier" will make identification quick and easy.

"Ghost Recon", my pet bot since 2019, became the champion in Robo Soccer segment of "Spectra 2.0" organized by Brac University Electrical and Electronic Club. It was the best performance from my bot since the day it was created. 

To build this bot, I incorporated an Arduino UNO, a HC-05 Bluetooth Module, two IBT-2 BTS Motor Drivers, four high torque 12V 300RPM motors, a 12V battery, one DP-Switch, connectors, connecting wires, and stainless steel for the chassis. It dimensions 25x25x18 cubic centimeters  and weights around 2.3KG.

What it does:

1. Runs face detection only on image datasets

2. Two detector options: Haar cascade (fast) or Res10 SSD (OpenCV DNN) (more robust)

3. Colab-ready notebook with step-by-step cells

4. Batch processing of ZIP/TGZ datasets; saves annotated images to outputs/ and optional face crops to crops/

5. Tunable parameters (Haar: scaleFactor, minNeighbors; DNN: confidence threshold)
   
   

A This project demonstrates a complete end-to-end IoT solution with a focus on embedded systems and industry practices:

• ESP32 Firmware (Arduino/PlatformIO) – MQTT-based device control with simulated temperature & humidity telemetry.

• Cloud Messaging (MQTT) – clean topic contracts (home/<device>/...) and JSON payloads.

• Backend – Dockerized Mosquitto broker and Node-RED Dashboard for real-time monitoring and control.

• Wokwi Simulation – fully testable without hardware, ideal for quick prototyping.

•  CI/CD – GitHub Actions workflow for automated firmware builds

This project reflects my interest in embedded IoT systems, cloud integration, and automation workflows — while keeping the implementation clear and extensible for future features like TLS security, Home Assistant integration, or additional devices.

Academic Freelancing Project taken as an Instructor. The goal of this project was to construct a Smart Window Shield.

I combined one UV Sensor, one Rain Sensor, one GSM Module, one Arduino Uno, one 16x2 LCD Display, two servo motors, and one DHT 11 Temperature and Humidity Sensor in this project. Under specific instances, the system will automatically open or close the window shield to protect residents, advise house owners about the room temperature, and contact SOS if an emergency happens.

This is an university project with my classmates for EEE-301, Digital Logic Design. The goal of this project is to identify COVID-19 suspects and recommend that they undergo COVID-19 testing.

This system was entirely built in analog circuitry. In this case, we constructed four, 7485 Comparator IC. There are two Comparator IC for Oxygen Level Saturation and two more for Forehead Temperature. We set one 7485 IC as the reference value and compare it to another 7485 IC. Follow the same steps to obtain the other objective.

When a person go through the gate, the results of the child circuits will go through a NAND Gate. We utilized a logic probe to identify the people who were present. Both findings will be deployed as inputs to the 4026 Seven Segment Counter IC, which will display the amount of suspects and notify the authorities for further surveillance.

This is a academic project with my fellow mates under EEE-305, Control System Course.

Without control systems, Electrical and Electronic Engineering [EEE] is incomplete. A system may be programmed to provide certain outputs based on a variety of factors. Developing this Project,  w e incorporated a MCU based Automated Temperature Control Box in this project. The box will operate the entire system under specific circumstances. It disconnects the system from the main grid during high temperatures and functions simultaneously during low temperatures.

A sensor integration based Do It Yourself (DIY) Project. It was a very inspiring project.

I was getting bored on a random day during COVID-19 lockdown. I glanced into my Sensor box and discovered an analog sound sensor LM386. So I constructed the project with the components I had at the time. The experiment resulted in my Static RGB LED Strip becoming a reactive Strip based on melody and harmony, which I find quite exhilarating while listening to music at night in a dark environment.

Academic Freelancing Project taken as an Instructor. The purpose of this project was to build a kitchen safety and security system with notification alarm. 

It was an ESP8266-based smart kitchen with automation and an ESP-DASH monitoring system. The principal objective of this project is to create a web-based prototype of an ESP8266-based smart kitchen monitoring and automation system. Several sensors, relays, and NodeMCU ESP8266 boards are used in the system. On the online dashboard, we can view all sensor data visually. The webserver can also send commands to operate kitchen appliances.