Let me introduce myself!
I am a graduate student at Georgia Tech. Most recently, I was a Machine Learning Intern at Apple Inc., where I designed an NLP pipeline utilizing transformer-based LLMs. Previously, I worked at StellarDNN, a lab at Harvard University, researching data augmentation techniques for astronomical datasets using generative networks.
I have demonstrated expertise in machine learning, data science and data analytics. I have worked with tabular, image, audio, text, and time series data and would love to collaborate on any such project!
Beyond tech, I love a game of tennis, a good book and I'm a student of classical dance.
EXPERIENCE
May 2024 - Aug 2023
Machine Learning Intern
Aug 2023 - Present
MS Computational Science
Aug 2021 - Aug 2023
Research Assistant
Jun 2020 - Aug 2023
Teaching and Research Fellow
Aug 2020 - Aug 2021
Data Analyst
Jan 2020 - May 2020
Data Analytics Intern
Jul 2019 - Aug 2019
ML and AI Intern
Apple Inc.
-
Designed a natural language processing pipeline leveraging transformer-based LLMs to achieve a 4x increase in action rate for objectionable content detection within Apple Music's lyrics corpus, resulting in a 55% reduction in manual review time
-
Implemented a theme categorization algorithm using topic modeling techniques to classify diverse content into distinct thematic categories precisely.
Georgia Institute of Technology
-
GPA: 4.0/4.0
-
Relevant Courses: Computer Vision, Data & Visual Analytics, Scientific Machine Learning, Computational Data Analysis
-
Graduate Teaching Assistant: CSE 6242: Data and Visual Analytics
-
Graduate Research Assistant: ViTAL Lab
StellarDNN, Harvard University
-
PI: Dr. Pavlos Protopapas
-
Published a research paper on a controllable disentangled generative adversarial network (GAN) to generate high definition images of the M87* black hole using observation data released by the Event Horizon Telescope Collaboration leveraging computer vision and other deep learning techniques.
-
Created an auxiliary classifier GAN to generate astronomical time series data in order to improve periodic star classification through data augmentation.
Univ.AI
-
Developed the curriculum for a year-long machine learning and data science course covering a range of topics including multi-layer perceptrons, convolutional neural networks, exploratory data analysis, tree and ensemble models, recommendation engines and generative modeling.
-
Taught labs and exercises to 800+ students for over 5 courses.
-
Led a team to establish and develop the curriculum for Futureschool.AI, a sister company offering cutting-edge AI/ML courses to high school students.
-
90% of students rated above 4.8/5.0 in course evaluation.
Schneider Electric
-
Designed and implemented user interfaces, backend technologies and custom web applications on the Salesforce platform using JavaScript, Java and HTML/CSS.
-
Integrated third-party platforms into Salesforce using native REST APIs.
-
Designed and implemented automated code smell and bug identification and removal toolset. Led to better code quality and reduced redundant code by 15%.
Schneider Electric
-
Led a team to design and build an image processing tool using convolutional neural networks to automatically pre-process and categorize product images, greatly reducing manual work across multiple sales/product teams.
-
Designed and developed a deep analytics dashboard to gain insights into support tickets logged by customers to increase development efficiencies. The dashboard was developed on the Einstein Analytics platform.
-
Awarded superlative grade for outstanding performance during the internship.
TietoEvry
-
Implemented neural networks and boosted regression tree models to analyze and forecast energy consumption in households.
-
Developed energy consumption models and implemented the analytics platform to proactively enable predictive maintenance of energy infrastructure.
PROJECTS
SoundSensei: Customizable Playlist Curation and Safe Listening for All
We present SoundSensei, an innovative music recommendation system, transcends the conventional boundaries of music curation by placing transparency, customization, and control at its core.
Using recommendation systems with Tensorflow/Keras