Expected December 2026
Master of Science, Artificial Intelligence
Johns Hopkins University, Whiting School of Engineering
Hey there 👋
I am Yosef Miller,
an AI Masters Student
and a Professional
Full-Stack Engineer.
Select Language:
importtorch,torch.nnasnnclassAIEngineer(nn.Module):def__init__(self):super().__init__()self.name="Yosef Miller"self.skills=["PyTorch", "Deep Learning",
"Computer Vision", "RL",
"Torchvision", "Transformers"]self.model=nn.Sequential(nn.Linear(len(self.skills), 16),nn.ReLU(),nn.Linear(16, 1),nn.Sigmoid())defforward(self, x):hireable_score=self.model(x)returnhireable_score
model=AIEngineer()x=torch.ones(len(model.skills))print("Hireable Score:", model(x).item())ABOUT ME
Who am I?
I’m Yosef Miller, a passionate problem solver pursuing a Master’s degree in Artificial Intelligence from Johns Hopkins University, with a diverse background that combines technical expertise with creative thinking. As a full-stack developer, I’ve led projects ranging from building an Uber-like platform for lifeguards to designing a complex SMS communication system and automating retail data workflows. With skills in AI/ML, cloud deployment, and an array of languages, I thrive on crafting scalable, user-friendly solutions. Beyond development, I enjoy mentoring, leading community outreach programs, and contributing to organizational projects to make an impact both in and out of tech.
Education
August 2024
Bachelor and Master of Arts, Talmudic Law
Ner Israel Rabbinical College
Research
Academic research projects in Machine Learning, Computer Vision, and Natural Language Processing conducted at Johns Hopkins University
Multivariate Time-Series Classification for Social Engagement Fraud Detection
Johns Hopkins University•Summer 2026
Time-Series ClassificationFraud DetectionMachine LearningSocial Media Analytics
This project evaluates supervised multivariate time-series classification methods for detecting social engagement fraud in short-form social media submissions. Each post is represented by seven days o...
Instagram Precision-Recall Curve
Comparing Reinforcement Learning Algorithms Using a Custom Version of Lunar Lander
Johns Hopkins University•Spring 2026
Reinforcement LearningDeep Q-NetworksTemporal Difference Learning
Four different RL algorithms are trained and evaluated within a custom version of Lunar Lander. Specifically, this study compares SARSA, Q-Learning, SARSA-λ with linear function approximation, and DQN...
DQN Landing Demo
Automatic Speech Recognition for Code-Switching
Johns Hopkins University•Fall 2025
Speech RecognitionCode-SwitchingTransfer LearningMultilingual NLP
Code-switching is a phenomenon in speech where speakers alternate between two or more languages or language varieties within a single utterance. Accurate transcription of code-switched speech remains ...
Autonomous Driving Domain Adaptation
Johns Hopkins University•Summer 2025
Domain AdaptationComputer VisionAutonomous DrivingTransfer Learning
Image classification models often experience significant performance degradation when deployed under conditions different from their training environment. This work addresses the challenge of domain s...
BDD100K Samples with Segmentation
Skills
Experience
PROJECTS
Differential-Drive Maze-Navigating Robot
constproject={name:'Differential-Drive Maze-Navigating Robot',tools: ['ROS2', 'Gazebo', 'Python', 'C++', 'Arduino', 'Docker],myRole:Johns Hopkins University Robotics Course Project,description: Developed embedded control software for an autonomous differential-drive robot to navigate a physical maze using ROS-based control, kinematics, and odometry. Integrated sensor feedback loops for obstacle avoidance and wall-following, achieving stable real-time path tracking on hardware.};Lifeguarding Agency Web App
constproject={name:'Lifeguarding Agency Web App',tools: ['Laravel', 'FilamentPHP', 'Livewire', 'Tailwind CSS', 'Ubuntu', 'PostgreSQL', 'Fly.io', 'DNS', 'Composer', 'AWS S3', 'Stripe', 'Sentry', 'Twilio', 'SMTP', 'Google Maps', 'Google Analytics', 'ReCAPTCHA],myRole:Founding Engineer & Lead Full Stack Developer,description: Lead developer for an Uber-like business startup that connects camps and pool owners with local available lifeguards. Fully automated with scheduled reminders and various user dashboards, accessible based on role. Integrated SMS notifications and a dynamic IVR using Twilio API to streamline the provided services. Coded in Laravel and FilamentPHP, integrated with numerous services, and deployed scalable Ubuntu and PostreSQL servers using Fly.io infrastructure.};Workflow Automation
constproject={name:'Workflow Automation',tools: ['Power Query', 'Shopify', 'EDI', 'Excel', 'VBA', 'Google Apps Script],myRole:Independent Contractor,description: Developed customized algorithms to automate the batch ingestion of retail invoice data for a supply distribution center. Utilized Advanced Power Query Editor as an ETL to extract relevant content from an EDI output in preparation for shipping document generation. Enhanced Shopify Online Store functionality through theme customization to improve user experience and support external vendor integrations.};SMS Communication System
constproject={name:'SMS Communication System',tools: ['Twilio', 'Laravel', 'MySQL', 'HTML', 'CSS', 'AlpineJS', 'Tailwind CSS],myRole:Lead Full Stack Developer,description: Designed and implemented a complex SMS communication system using Twilio API to facilitate communication between parents and volunteers nationally. Developed a custom CRM to manage client data and track communication history.};Audio Lecture Distribution System
constproject={name:'Audio Lecture Distribution System',tools: ['Laravel', 'MySQL', 'HTML', 'Responsive Design', 'Redis', 'AWS S3', 'Multi-Tenant Architecture', 'Eloquent ORM],myRole:Full Stack Developer,description: Developed a full-stack web application to efficiently query a multi-tenant SQL database using a versatile taxonomical filtering scheme. Integrated an object-oriented pattern with Laravel PHP framework to composite request data into a query structure to provide advanced query capabilities and search methods.};Optical Character Recognition
constproject={name:'Optical Character Recognition',tools: ['MATLAB', 'JavaScript', 'Canny Edge Detection', 'Shape Context Descriptors', 'Template Matching', 'L2 Norms', 'Statistical Comparisons],myRole:Science Fair 1st Place Project,description: Developed my own OCR algorithm to isolate and extract text within an image. Implemented Canny Edge Detection and Shape Context descriptors for improved pattern recognition. Provided a GUI to train and detect loaded shapes, including plots of matching results. Micro-optimized the MATLAB code for time-efficient execution by utilizing matrix operations. In addition, wrote a JavaScript implementation of a Template Matching approach using L2 Norms and various statistical comparisons.};