Rubrik, Inc.

SWE Intern | May 2018 - Aug 2018

• Designed and implemented a robust database system with efficient random reads and sequential scanning, used to store email metadata in the cloud.
• Also added universal compatibility with various cloud providers (S3, Azure, Google Cloud) and further augmented it with Zstd compression and file deduplication.

Brilliant

Data Science / Education Intern | May 2017 - Aug 2017

• Conducted data analysis using SQL and Python on 1+ million emails to evaluate email algorithm effectiveness.
• Designed and produced curricula for "Math for Computer Science" and "Intro to Data Science" courses.

UC Berkeley Data 100

Head Teaching Assistant | Jan 2018 - Present

• Prepared materials, taught lab section, held offce hours, and developed assignments.
• Facilitated class logistics, including managing staff, organizing exams, and overseeing enrollment for a class of 1400 students.
[Teaching Assistant | Aug 2017 - Dec 2017]

UC Berkeley RISELab

Research Assistant | February 2018 - Present

• Applied convolutional neural networks to detect and classify arrhythmias from electrocardiogram readings.
• Also built a hierarchical classifier that parses and analyzes data from doctors' notes.

BerkeleyTime

Data Scientist/Product Manager | Jan 2018 - Present

• BerkeleyTime is an augmented course catalog that provides data on courses, enrollment trends, grade distributions, and more. I serve as the Product Manager and Data Scientist for BerkeleyTime, and was previously Lead Engineer.
• We just wrapped up a course scheduler, and are currently working on a data-centered initiative with student course data (such as course recommendations, automatic course plan generation, and intelligence course classiffication).

Neural Net Applications | Python

• Implemented a neural net from scratch (creating a graph class, implementing matrix operations, backpropagation, etc).
• Applied NNs to solve various problems, such as digit classification, language classification, and function approximation.

Song Classifier | Python, TensorFlow

• Created a deep learning classifier that categorizes songs as either country or hip-hop based on lyrics frequencies.
• Achieved an accuracy rate of 89% and was selected as one of 10 winners of a 200-member Kaggle contest.

Jesture | HTML, CSS, JS, jQuery, Python, C, AppleScript

• Created gesture detection application with the Synaptics touchpad; implemented API for Spotify, Slack, Facebook, and more, then linked to a sleek web UI.