Advanced Machine Learning for Credit Card Fraud Detection
Conducted a fraud detection analysis on over 1 million transactions using KNN and Logistic Regression, employing Python's sci-kit-learn for model building and Flask for deployment. The project involved normalizing data with StandardScaler, utilizing pandas for data manipulation, and matplotlib and seaborn for visualization, with StratifiedKFold enhancing model evaluation, focusing on accuracy and ROC-AUC scores to refine detection techniques.
Advise-A-Gator
Crafted an interactive visual roadmap and a dashboard by streamlining course data into JSON format, reducing redundancy by 50%, enhancing data consistency across platforms, and Implementing continuous integration and pre-commit hooks. Led a 3-member team to improve app efficiency by 30 percent using React and Node.js, effectively delegating tasks according to the strengths of group members.
Interactive Jeopardy Quiz Platform
Integrating a Raspberry Pi Pico with a joystick and RGB matrix screen, crafting a dynamic Jeopardy-style quiz board that supports real-time user interactions. Programming joystick inputs and RGB display outputs using MicroPython, building a robust game system architecture that showcases innovative solutions and complex problem-solving in IoT. Engineering a responsive quiz platform with interactive LED feedback and optimized user interfaces on a Raspberry Pi Pico. My technical approach included detailed circuit design and efficient code for signal processing and display management, significantly enhancing user engagement and deepening learning outcomes.
VueJS Personal Portfolio
Developed and deployed a personal portfolio website, and engineered a dynamic portfolio using Vue.js, JavaScript, and CSS, enhancing user interface and experience through reactive components and responsive design. Implemented advanced features like live content updates and cross-browser compatibility to showcase technical projects and skills effectively. Streamlined development with best practices, utilized Git for robust version control and GitHub Actions for automated CI/CD pipelines, significantly reducing deployment times and manual errors. Hosted on Netlify for high availability and performance, demonstrating scalability and professional branding.
Environment Dashboard
Designed and executed a secure, MQTT-based communication system on a Raspberry Pi Pico, incorporating advanced SSL/TLS encryption to ensure data integrity and confidentiality within a cloud environment. Employed HiveMQ's cloud services to achieve scalable, real-time data transfer and monitoring. Developed an integrated dual-sensor system, utilizing a high-precision BME280 sensor alongside Pico’s onboard temperature sensor to capture and compare diverse environmental data. This dual approach enhanced data reliability and provided a comprehensive environmental analysis. Engineered a Flask-based web dashboard, which displays real-time environmental data through interactive gauges and dynamic charts using Socket.IO for live updates. This setup enabled immediate data visualization and facilitated real-time decision-making in environmental monitoring. Leveraged HiveMQ's cloud platform to manage IoT device connectivity and data handling, ensuring robust data publishing and efficient handling of subscriptions. Configured a cloud-based dashboard to offer instantaneous insights into temperature, humidity, pressure, and other sensor readings, demonstrating proficiency in IoT network management and cloud integration.
Scrap or Spec: Acutec Data Dive
Utilized SQL to perform a detailed analysis of over 1 million records, pinpointing 'RED' part defects. This rigorous analysis significantly enhanced production efficiency and reduced waste in aerospace part manufacturing. Conducted deep analysis of nonconformance issues affecting aerospace orders worth over $221 million. Leveraged advanced data visualization techniques to generate actionable insights, driving strategic improvements and optimizing material utilization. These efforts notably increased production profitability and operational efficiency.
Vegetable Oil-Obesity Correlation Study
Leveraged R to conduct detailed analysis on the correlation between vegetable oil consumption and obesity rates, utilizing over 1 million data points. Applied data cleaning and manipulation techniques using dplyr and tidyverse to prepare datasets for analysis. Employed linear regression models to identify and quantify correlations between dietary factors and health outcomes, focusing on the impact of vegetable oil on obesity rates. Results contributed to a deeper understanding of dietary impacts on public health. Created advanced visualizations using ggplot2 to illustrate the relationship between vegetable oil availability, diet affordability, and obesity rates. These visualizations were pivotal in informing discussions on nutritional strategy and public health policies.