Electra Genius

Single-Cycle RISC-V Processor using SystemVerilog on a Nexys A7 (Artix-7) FPGA. Project includes complete datapath and control logic with instruction memory, data memory, ALU, immediate generator, and branch comparator. It supports the complete RV32I instruction set (R, I, S, B, U, J types).

A 5-Stage Pipelined RISC-V Processor designed and implemented on FPGA (Artix-7 Nexys A7). Supports RV32I instructions set (R, I, S, B, U, J types) with ALU, control unit, hazard detection, forwarding, and pipeline registers. Verified through simulation and hardware testing with optimized timing and 4× performance gain.

People share opinions on Twitter every second. Companies, governments, and researchers want to know what people feel about products, events, or topics in real time. Reading tweets manually is impossible at scale. SentimentFlow is a deep learning based sentiment analysis project for tweets. It trains two models like a basic RNN and an LSTM network.

A Smart Anti-Theft Car Security System implemented on FPGA to detect and prevent unauthorized access. The system uses real-time monitoring and control logic to enhance vehicle safety and response.

Config files for my GitHub profile.

I developed a compact, autonomous robot that follows a human using Arduino and IR sensors. This innovative solution has applications in personal assistance, security, , Warehousing and Logistics, Retail and Hospitality and interactive robotics

I developed the IoT-Based Electricity Theft Detection system, which uses IoT sensors and real-time analytics to identify and prevent electricity theft. The system features anomaly detection, remote monitoring, and automated alerts for secure energy distribution.

This is my first repository and it is based on object detection and object tracking of vehicles. It will first detect and count the vehicles.

This repository features the design and Proteus simulation of an AM transmitter and a Superheterodyne AM receiver for the medium-wave band. It includes MOSFET-based amplitude modulation, frequency down-conversion to 455 kHz IF, and signal recovery via envelope detection using basic electronic components.

This system measures heart rate, temperature, and SpO2 levels using real-time sensors. Includes fall detection via MPU sensor—ideal for elderly or patient care. Health data is shown on a live dashboard, enabling remote monitoring, early alerts, and smart health management.

PID-based one-axis ball balancing robot using Arduino. Uses an ultrasonic sensor and servo motor to stabilize a ball at a target distance. Real-time tuning of Kp, Ki, Kd via Serial. Includes stability detection and automatic neutral reset. The system stabilizes within 30s using empirically tuned parameters.

I developed the Egg Incubator Project using an Arduino to maintain optimal hatching conditions with temperature and humidity control, and periodic egg turning. With a 100% hatch rate, this incubator features a DHT11 sensor, stepper motor for egg turning, LCD display for status, and buzzer alerts for critical conditions.

An IoT-powered system for real-time air quality monitoring and analysis. This project integrates environmental sensors with a machine learning model to predict and assess air quality indices. Features include data visualization, predictive analytics, and automated alerts for actionable insights.

Implementation and analysis of cache replacement policies (Random and Least Recently Used) in a C++-based cache simulator. This project explores cache architecture behavior, evaluates eviction strategies, and measures performance metrics such as cache hits, misses, and flush counts.

A machine learning project for binary classification of skin cancer as malignant or benign, utilizing models like XGBoost, LGBM Classifier, Adaboost, SVM, and Logistic Regression. Features comprehensive data preprocessing, model training, and evaluation for accurate diagnosis.

An integrated MATLAB–ML system for early fault detection in induction motors. Detects six faults broken rotor bars, stator short, ground fault, overloading, eccentricity, and voltage imbalance using KNN and Decision Tree models for accurate, unified, and reliable predictive maintenance.

A real-time music genre classification system that combines classical machine learning techniques with a CNN trained on mel-spectrograms. The hybrid model utilizes DSP-based feature extraction and ensemble learning to achieve high accuracy and low latency for practical audio applications.

A hands-on workshop template that demonstrates how to orchestrate CrewAI agents for planning, research, writing, and review workflows. The stack combines CrewAI with LangChain tools, a FAISS-backed Retrieval-Augmented Generation (RAG) pipeline, and a Streamlit frontend.

This project focuses on segmenting retinal blood vessels from fundus images using a U Net based deep learning model. The goal is to build a simple, clear, and reproducible pipeline that loads the dataset, pairs images with their masks, trains a U Net, evaluates the model, and visualizes the predicted segmentation maps.

This project uses a U Net based deep learning model to perform pixel level semantic segmentation on CARLA simulator images. The goal is to classify each pixel of a driving scene into categories such as road, vehicles, pedestrians, buildings, vegetation, and background.

BreastNet is a neural network based model for early detection of breast cancer. The goal is to classify a given breast tissue sample as benign or malignant using numerical features extracted from medical tests. The project builds a full pipeline: data loading, preprocessing, model design, training, evaluation, and result visualization.

Recycling fails when different types of waste get mixed together. Manual sorting is slow and unhygienic, so a smart automated system is needed. RecycleVision is a deep learning based image classifier that identifies different types of waste and makes automatic segregation possible for smart bins and recycling plants.

Millions of transactions occur daily, and fraudulent activity can cause significant monetary loss. Manual inspection of transactions is impossible due to data scale and speed requirements. Build a machine learning pipeline that identifies fraudulent transactions accurately, even when the dataset is extremely imbalanced.

Wild mushrooms can be edible or poisonous. Many of them look very similar. For a normal person it is hard to tell which mushroom is safe and which is dangerous. The task is Given the physical features of a mushroom, predict if it is **edible** or **poisonous**. This is solved as a binary classification problem using Decision Tree and Random Forest.

This repository demonstrates a complete machine learning workflow for banknote authentication using a hand coded Linear SVM model. It includes data preparation, training loop design, vectorized computation of gradients, model metrics, and a visualization module based on two selected features.

This project builds a complete image classification system that identifies clothing items from the Fashion MNIST dataset. The goal is to train a CNN that can accurately classify outfits such as T-shirt, trouser, pullover, dress, coat, sandal, shirt, sneaker, bag, and ankle boot.

This project builds a high accuracy image classification system for the CIFAR100 dataset using the WideResNet28x10 architecture. The goal is to achieve strong top 1 accuracy through advanced augmentation, stable training strategy, and detailed model analysis.

Stereo Depth Estimation is a computer vision project that extracts depth information from a pair of images captured from two slightly different viewpoints. The goal is to understand how far objects are from the cameras by comparing the differences between the left and right images.

Designed and evaluated content-based, collaborative, and hybrid recommender systems for modern movies (2015–2025). The project involved API-based data collection, data scraping, feature engineering, TF-IDF text modeling, matrix factorization, and hybrid score fusion, demonstrating practical application of recommender system methodologies.

Developed a real-time smart energy monitoring system on STM32F746 using FreeRTOS, managing tasks for sensor acquisition, Vrms, Irms, and kWh computation, and local display. Integrated an ESP32 IoT interface to transmit processed data to a cloud dashboard, ensuring comprehensive remote tracking of electrical parameters and consumption.