Hello, I'm

My Projects

• Reinforcement Learning assisted Adaptive difficulty of Proof of Work (PoW) in Blockchain-enabled Federated Learning

  Developed a novel Reinforcement Learning agent using Deep Q-Learning (DQN) to enable adaptive difficulty of Proof of Work (PoW) in a private blockchain (distributed systems), ensuring system fairness (<30%) and enforcing a time constraint of 20 seconds using Python, node.JS and shell scripting.

• Dockerized API Web Service

  Architected a platform-independent Flask-based, Dockerized API to validate, process, and store user-provided JSON receipts. Introduced a unique ID system for efficient receipt data retrieval, elevating both system efficiency and user experience.

• Tokyo-Olympic Data Analytics using Azure

  Engineered a robust data analytics pipeline utilizing Azure tools: Data Factory, Databricks, Data Lake, Synapse Analytics, and Power BI.

• Feature Extraction and Statistical Analysis of Urban Mobility

  Improved traffic flow predictions for locations in London by 15% through the application of advanced machine learning models.

• Email Honeypot and Spam Detection

  Crafted an email honeypot to ensnare spammers (MX records, DNS). Collected data to create a real dataset, trained, and compared spam detection classifiers, including Multinomial Naive Bayes (ACC=98%), Bernoulli Naive Bayes (ACC=98%), and SVM (ACC=95%).

• Accurate PM2.5 Prediction using Low-Cost Sensor and Weather Dataset

  Accomplished precise PM2.5 prediction by leveraging low-quality sensor data and applying machine learning algorithms. Conducted a comprehensive evaluation of six models, surpassing EPA's baseline with a 15% increase in accuracy. Addressed cost constraints and highlighted the efficacy of low-cost sensors in air-quality monitoring.

• Covid Friendly Gesture Kiosk

  Devised and implemented a fully touchless kiosk desktop application, resulting in a significant 100% reduction in physical contact for food orders by integrating a camera sensor. Crafted, labeled and employed a dataset of 250 images to train the system.

• Implementation of Proprietary and Open Source 5G Architecture

  Successfully implemented and evaluated proprietary and open-source 5G architectures using Amari Callbox Pro, showcasing expertise in advanced networking technologies. Deployed 5G NSA and SA setups and demonstrated adaptability with SRS RAN for open-source integration.

• APSOC Using High-Level Synthesis

  Engineered a cutting-edge APSOC through FPGA programming with Vivado HLS using Zynq2000 - Applied advanced high-level synthesis techniques to optimize implementation, demonstrating proficiency in PS PL integration using the AXI interface.

• FPGA Implementation of AES Algorithm

  Successfully implemented the AES algorithm on a Kintex7 KC7325 FPGA using Vivado Design Suite, achieving robust and secure encryption. Possess expertise in FPGA programming, algorithm implementation, and seamless hardware-software integration.

• Home Automation System

  Engineered a streamlined Home Automation System utilizing Arduino, seamlessly integrating LEDs, IR controller, and relays for remote control of household appliances.

• React Portfolio Website

  Designed and implemented a Next.JS based portfolio website, showcasing expertise in React, Next.JS, and Tailwind CSS.

• AB Testing and Multi-Armed Bandit

  Designed and performed AB testing and Multi-Armed Bandit experiments to dscover the difference in the two distributions choosing the optimal one.