Hi, I'm Yonathan Daniel
Engineering Solutions, One Line of Code at a Time
About Me
As a software developer pursuing my Master of Science in Computer Science at Columbia University, I specialize in backend development and cloud technologies. With experience in AWS, machine learning, and full-stack development, I'm passionate about creating efficient and scalable solutions.
My background includes significant projects in healthcare AI, e-commerce, and autonomous systems, demonstrating my ability to tackle complex technical challenges and deliver innovative solutions.
Projects
Technologies used: TensorFlow, AWS (SageMaker, OpenSearch, API Gateway, Lambda), Websockets
- Developed a convolutional neural network achieving 76% accuracy in skin condition prediction
- Implemented real-time chat feature using API Gateway and websockets
- Utilized AWS services for enhanced functionality and scalability
Technologies used: Python, XGBoost, Jupyter, Machine Learning
- Implemented multiple ML models including XGBoost and CNN for phishing detection
- Developed feature extraction techniques for website classification
- Collaborated with team to analyze and process large datasets of website characteristics
- Created comprehensive documentation and analysis in Jupyter notebooks
Technologies used: AWS (Codepipeline, CloudFormation, Lambda), CI/CD, Infrastructure as Code
- Implemented picture uploads with label metadata and image search feature
- Incorporated CI/CD via AWS Codepipeline, decreasing setup time by 40%
- Delivered Infrastructure as Code using AWS CloudFormation for automated deployment
Technologies used: Python, Flask, PostgreSQL, Google Cloud Platform
- Built backend processing all store operations and transaction logging
- Developed vendor and product management features for store owners
- Designed and implemented a comprehensive database schema for products, orders, users, and vendors
- Deployed via Google Cloud Platform for improved scalability
Technologies used: ROS, Gazebo, Python
- Led development of maze design in the simulation
- Researched ROS Gazebo mechanics for realistic environment simulation
- Enhanced autonomous capabilities for rescue and reconnaissance missions
Technologies used: PyTorch, Distributed Data Parallel (DDP), CIFAR-10, Multi-GPU Training
- Implemented Distributed Data Parallel (DDP) for efficient multi-GPU training
- Trained ResNet-18 on the CIFAR-10 dataset, achieving competitive accuracy benchmarks
- Designed a modular code structure with separate components for training, model definition, and utilities
- Optimized the training pipeline for multi-GPU environments, reducing training time significantly
- Demonstrated proficiency in distributed deep learning and modern PyTorch practices