Hi, I'm Kevin Miller
I am a Data Scientist, Software Engineer, and Educator with a strong passion for solving meaningful problems at the intersection of data and technology, particularly in the healthcare domain. My work spans applied research, software development, and complex problem-solving across diverse industries. As a Training Consultant, I also partner with industry professionals to deliver practical, hands-on guidance in Data Analytics and Machine Learning, helping organizations address real-world challenges with effective, data-driven solutions.
CONTACT ME
Data Science & Analytics | Software Development | AI/ML
Machine Learning Projects
Diabetes Prediction (AI/ML)
This project explores the application of machine learning algorithms to predict an individual’s likelihood of developing diabetes, using data from the 2020 Behavioral Risk Factor Surveillance System (BRFSS). A number of traditional Machine Learning algorithms were evaluated along with a PyTorch model. A sample Flask application was also developed. The dataset, provided by the CDC, is publicly available [here].
Technologies Used:
Python, PyTorch, Google Cloud Run, Docker, Flask, Sklearn Visit GitHub Demo AppAn AI-Powered Decision Support System for Enhancing Optimal Drug Prescription
This applied research project presents an AI-powered decision support system designed to assist in the drug prescription process. It explores two primary approaches to address this challenge: a traditional machine learning method and a large language model (LLM) enhanced with Retrieval-Augmented Generation (RAG).
Technologies Used:
Python, LLM, RAG, Flask, Synthesia, MySQL Visit BlogUsing K-Means for USDA Food Project (AI/ML)
In this project, an exploratory analysis of USDA food data to cluster items based on their nutrient profiles was performed. The resulting groups largely aligned with common food categories (e.g., sugary foods, meats). Notably, lean meats clustered with seeds and legumes, suggesting nutritional similarities and supporting the idea that certain plant-based foods may serve as viable meat alternatives.
Technologies Used:
Python, Pandas, Sklearn, K-Means, PCA, t-SNE Visit GitHubPDF Embedding and Query System with LangChain and RAG (AI/ML)
This project demonstrates how to build a system for creating and querying embeddings from PDF documents using LangChain, Chroma, and OpenAI. The system loads PDF files, splits the text into chunks, creates embeddings, and stores them in a vector database. It also provides functionality to query these embeddings for similarity and generate responses.
Technologies Used:
Python, LangChain, Chroma, OpenAI, python-dotenv, shutil, argparse Visit GitHubFamily Face Tracking System (AI/ML)
The Family Face Recognition Tracking System is a Python-based application designed to detect and track faces in real-time using a webcam. This application leverages a foundational face recognition model trained on images of my family members. It identifies known individuals and saves any unknown faces to the filesystem for later review.
Technologies Used:
face_recognition, opencv-python, numpy, matplotlib, seaborn, scikit-learn, threading, argparse Visit GitHub Visit BlogTransfer Learning with PyTorch for classifying bird species (AI/ML)
This project utilizes transfer learning with EfficientNet-B0 for bird species classification. It employs PyTorch’s Dataset and DataLoader for efficient data handling and extends the EfficientNet-B0 model to create a derived model. A training loop was implemented to monitor training and validation losses, and the model's accuracy was assessed on a test set after training.
Technologies Used:
PyTorch, PyTorch Image Model(timm), Pandas, Numpy, Matplotlib Visit GitHub Visit BlogWeb Development Projects
KWIZZ (Python)
Kwizz is an innovative web application designed to enable users to create, host, and participate in competitive quizzes. Kwizz offers a dynamic platform where quiz creators can craft engaging questions and participants can compete in real-time to see who can answer the questions the fastest. Kwizz makes quizzing fun and interactive.
Technologies Used:
HTML, JavaScript, HTMX, Socket.IO, CSS, Bootstrap 5, Python, Flask, Flask-SocketIO, Redis, MariaDB, SendGrid API, Pytest, Docker, gevent, Gunicorn, Docker Compose, Caddy, GitHub, VPS, GHCR, GitHub Actions ScreenshotsMyVideoRama (Go)
MyVideoRama enables users to craft personalized study programs using freely available YouTube videos. Users can create courses, seamlessly add videos to their curriculum, and annotate videos with study notes. This platform empowers learners to structure their learning journey efficiently, harnessing a diverse range of free educational content from YouTube.
Technologies Used:
HTML, JavaScript, HTMX, CSS and Bootstrap 5, Go Standard Library, Redis, MariaDB, SendGrid API, Go Standard, Testing Package, Docker, Docker Compose, Caddy, GitHub, VPS, GHCR, GitHub Actions ScreenshotsNyamUP (Python)
NyamUP is designed to help users manage meals, nutrients, and exercises using data from the Canadian Nutrient File. It provides an intuitive interface for tracking nutritional intake and physical activities, offering a comprehensive solution for maintaining a healthy lifestyle. The application uses modern web tools for a responsive, user-friendly experience.