I recently graduated from UC Irvine with a BS in Computer Science. I specialized in Artificial Intelligence and minored in Statistics. I love to work on projects in Data Science, Machine Learning, and Web Development. My favorite languages to code in are Python and C++. I am currently seeking a Software Engineering position in the Bay Area.
An analysis of supervised learning using different classifiers, word embeddings, and neural net architectures on 3 labeled datasets from academic studies on fraudelent review detection. I represent the reviews usings BOW, Doc2Vec, and GloVe. I train classifiers such as logistic regression and a LSTM neural network on the reviews. I evaluate the models using accuracy and AUC. After thorough experimentation, I achieved models that are more than 90% accurate on test data. I programmed in Python for this project, using the sci-kit learn and torchtext libraries to train classifiers and matplotlib to create graphs.
I implemented multiple common graph algorithms in C++. I wrote Bash scripts to efficiently collect large amounts of data on these algorithms and then analyzed the results in Jupyter Notebooks. I implemented breadth first search and algorithms to get the diameter, clustering coefficient, and degree distribution all in optimal run times. I used Python and matplotlib to plot the data.
During my summer at the Bosch Research and Technology Center, I worked on a next generation speech dialogue system called the Ciara SmartHome System. This software goes in all Bosch appliances and allows users to perform tasks like starting their Bosch dishwasher or oven simply by speaking to their device. My impact on the project was to use Python and Flask to create the dialogue engine component of this software. A statistical learning algorithm performs sentiment analysis to extract an intent from the user’s voice command. I created a dialogue engine that maintains a finite state machine of the state of each appliance and responds to each intent by either completing the task, asking followup questions, or reporting an error. I tracked the finite state machine by reading and writing JSON to a MySQL database, using the Python MySQL library. At the conclusion I had a fully functional demo working in Chinese and English using a web UI and a simulation of home virtual appliances. From this experience, I learned how a speech dialogue pipeline works and refined my backend web development and natural language processing skills.
While working as a summer intern for Cadence Design Systems, I built a chatbot to streamline the process of employees getting answers to their HR questions. I started by studying the most common questions employees had. These ranged from questions about healthcare, immigration, and 401ks. I then studied all the common responses to these questions. Next, I created a chatbot using Node.js, Azure Bot Service, Microsoft LUIS, and Microsoft QnA Maker. Micosoft LUIS allowed me to match HR related questions with an intent and using QnA Maker, I created a knowledge base of intents and helpful responses for those questions. I used HTML and CSS to customize the UI of the chatbot and make the appearance more appealing. By the end of the summer, employees were using the chatbot and, in turn, lightening the heavy load of the HR department.