The CS Foundation
Like most freshmen at NIT, I followed the herd — diving into web dev and DSA because it was the thing to do. Those projects taught me how to build, sure. But they also sparked a deeper question: Is this really what I want to build? Turns out, my real curiosity was waiting in circuits, signals, and the invisible world of wireless.
ChinGPT
ChinGPT — as I like to call it — is a web application powered by OpenAI’s GPT-3.5-Turbo. I built it with the vision of creating an AI assistant that anyone could access, using phone number registration and OTP-based user verification. Users can interact with the assistant through both text and voice input. Originally designed as a smart assistant for farmers, the project even drew a $75 offer while I was building it — but I chose to open-source it instead, keeping it free and accessible on GitHub.
Peeking Emissions Extension
A custom web browser extension to measure the carbon footprint of users as they surf the net. In addition to the extension, we made a website to rank different web pages based on their eco-friendliness as well.
Hardware & Machine Learning
Bridging the gap between the physical and digital, I explored everything from microcontrollers to deploying machine learning models on edge devices.
Arduino Room Environment Monitor
My first hands-on project in ECE. A simple yet fundamental system using an Arduino to display real-time room humidity and temperature on an LCD screen.
Warehouse Automation Bot
A Digital Systems Design project involving a bot for RFID-based cargo retrieval. I developed the state diagrams, K-Maps, custom PCB schematics, and implemented the MIMO controller in Verilog.
Solar Potential Estimation Tool
A web-based tool analyzing satellite images to calculate solar potential. It generates heatmaps on buildings and estimates power output. The basis for my IEEE research paper.
Non-Contact Soil Moisture Sensor
Co-authored a technical paper on a system that uses a webcam and an SVM model deployed on a Raspberry Pi to determine soil moisture levels (dry, optimum, saturated) from an image.
Deep Dive into Research
Collaborating with esteemed institutions, I delved into advanced topics like satellite communications and signal processing, contributing to cutting-edge research.
Indigenous GPS Receiver
A project in collaboration with the Indian Army Signal Corps. I captured and processed raw live GNSS signals using RTL-SDR and GNSS-SDR, comparing results with our in-house MATLAB scripts.
DOA Estimation @ IIT Madras
My largest ongoing project with the TelWise group. I developed a FFT-based "beamspace" MUSIC algorithm, achieving an 8x speed improvement over traditional variants for Direction of Arrival estimation.
Publications and Conferences
Comparative Analysis of UNet and UNet++ for Solar Potential Estimation
✔️ Accepted & Presenting
INSPECT 2025, IEEE Intl' Conference @ IIITM Gwalior (7–8 Nov 2025)
This paper evaluates the performance of UNet and its successor, UNet++, for semantic segmentation of satellite imagery to accurately estimate the solar energy potential of urban rooftops.
Non-Contact Based Soil Moisture Sensing using ML on Edge
👥 Co-Authored
Technical Paper
We present a cost-effective, non-contact method for soil moisture detection by deploying a Support Vector Machine (SVM) model on a Raspberry Pi, which classifies soil images captured by a webcam.