• Implemented machine learning models with Keras, Tensorflow, and XGBoost along with Optuna hyperparameter optimization for multiple regression of a synthetic dataset of over one million samples.• Employed data visualization tools including Matplotlib, Seaborn, Plotly, and Pandas to illustrate the performance of trained regression machine learning models, demonstrating their feasibility with Jupyter Notebook.• Designed and implemented ETL workflows using Xarray, Dask, and GeoPandas to process GeoJSON andNetCDF data, helping manage restrictions of real-world airspace based on flight conditions.• Managed a SQLite database to store airspace information and created queries to update it from a separate REST API.• Independently developed a C# Unity environment and Python backend utilizing NumPy and WebSockets,deployed using Docker and AWS EC2 with S3 to simulate dense air spaces of UAVs.• Developed an automated Python testing suite to unit test simulation parameters and ensure proper initialization of the simulation environment.

• Spearheaded data science projects as a Research Lead, collaborating with a team of researchers to utilize cloud computing and deep learning with geospatial imagery for remote sensing and conservation of mangroves.• Experimented with state-of-the-art deep-learning segmentation algorithms such as DeeplabV3 with PyTorch to segment aerial imagery.• Developed a classification and visualization dashboard using Dash, Plotly, and Xarray, incorporating Scikit-learn for automated data acquisition and classification to track long-term trends in mangrove extent.• Applied tree-based ensemble models and feature engineering techniques with LightGBM and Random Forest to classify satellite imagery, deploying on AWS Sagemaker.• Developed a CNN-MLP hybrid deep-learning image classification network with TensorFlow and Keras to detect mangrove areas using geospatial imagery data fusion, achieving an accuracy of 95%.• Produced machine learning reference datasets for mangrove remote sensing applications, leveraging NumPy and Dask to format labels and imagery in HDF5 format.

• Created curriculum and instructed students for an undergraduate course on subjects including introductions to Python, OpenCV, and Laser Cutting• Assisted students with group class projects involving machine learning, computer vision, and mechanical design

• Implemented and tested RNN and Transformer-based classifiers such as BERT in Tensorflow and Keras with NLTK to classify incoming IT emails using fastText embeddings. The final LSTM model was deployed on AWS Lambda, resulting in a 53% reduction in the manual sorting of new tickets.• Created an NLP Sentiment Analysis tool using Scikit-learn and Word2Vec embeddings to classify the sentiment of over 1000 internal social media posts daily, inserting the sentiment into a PostgreSQL server for proactive monitoring of staff complaints.

• Implemented a Convolutional Neural Network utilizing the InceptionV3 architecture to classify mangrove UAV imagery to achieve 93% accuracy• Implemented SVM and K-means baselines for mangrove image classification and clustering using Scikit-learn• Utilized ArcGIS and Agisoft Metashape to generate imagery from raw drone images for Deep Learning tasks