Mary Troiani
CS Masters Student with a passion for Artificial Intelligence
About Me:
As a cognitive science student at Case Western Reserve University I was always fascinated with the brain and cognition. I consider the mystery of how the brain affords intelligence one of the most important questions of our time, and more and more I’m convinced that this mystery will be solved through advances in artificial intelligence. In 2015 I quit my job to teach myself how to program, and now I love taking what I’ve learned about AI to create smart applications. I’m currently refining my skills as a Georgia Tech computer science masters student specializing in interactive intelligence, and I’m looking for opportunities to apply my knowledge to industry. Please feel free to get in touch here.
Skills
Languages:
Tools:
Data Science and ML Libraries: Pandas, NumPy, Scikit-learn, Keras, TensorFlow, OpenCV, SciPy, Matplotlib, Seaborn, NLTK
Projects
Deep Learning Dog Breed Prediction
Uses CNNs and transfer learning from RESNET50 with Keras and Tensorflow to predict the breed of a dog from a photo with over 80% accuracy.
Projects
Train a Smartcab to Drive
Uses reinforcement learning and Python to direct a virtual car to navigate to a destination while obeying traffic lights and US right-of-way laws.
Projects
Sign Language Identifier
Uses Hidden Markov Models to recognize sequences in gestures in American Sign Language from pre-processed videos to predict the English equivalent.
Projects
Customer Segmentation
Using PCA and K-means algorithm create customer clusters to determine vendor type based on purchasing habits.
Projects
Identifying Donors
Using Logistic Regession, Gradient Boosting Classifier, and Random Forest Classifier group potential donors into those most likely to donate.
Projects
Isolation Game-Playing Agent
Agent using minimax and alphabeta pruning search with a custom heuristic scoring function that plays isolation where pieces move like knights.
Projects
Sudoku Solver
Solves sudoku puzzles in Python using constraint propagation, including using the naked twins strategy.
Projects
Unbeatable Tic Tac Toe
Tic Tac Toe player that cannot be beaten. Uses the minimax algorithm in JavaScript to choose optimal moves.
Data Visualization Projects
Mapping Roller Derby Across the World
CSS and D3.js
Data Visualization Projects
Doping in Pro Bike Racing Scatterplot
CSS and D3.js
Data Visualization Projects
Mapping Meteorites Across the Globe
CSS and D3.js
Data Visualization Projects
Global Monthly Temperature History
CSS and D3.js
Data Visualization Projects
National Contiguity Force Directed Graph
CSS and D3.js