Close

Aditya Varshney

EECS at UC Berkeley | Data Science Enthusiast | Student Leader | CS Mentor

Download Resume

Welcome!

I am an EECS student in the UC Berkeley College of Engineering, Class of 2020. I enjoy learning about technology and applying my technical and leadership skills to solve real-world problems.


I'm particularly interested in machine learning, data science, and software design for fields like security, education, and e-commerce at scale. Take a look at some of my work so far!

Experience

Apple

Information Security Intern - Data Science

  • Used graph-based learning to speed up threat response and vulnerability management investigations into Apple's internal network
  • Improved asset management by incorporating state-of-the-art community detection, role extraction, and anomaly detection techniques to provide a stronger understanding of the different components within the internal network
  • Integrated graph algorithm outputs and analysis with the Beagle library for smooth and easy graph visualization and visual investigation
  • Tools used included Python for back-end data analysis and React and Flask for the front-end
  • Presented project and findings to Apple security executives and upper AIS management

Linc Global

Machine Learning and Software Engineering Intern

  • Integrated sentiment analysis into the company chatbot platform using company data
  • Implemented key product feature extraction using NLP libraries (spacy, NLTK, wordnet, and more)
  • Optimized chat bot product recognition using Knapsack
  • Helped design and implement the REST API framework to enable smarter conversations

Snap-On Tools

Algorithms and Decryption Intern

  • Worked on the Applications team at Snap-On Tools
  • Found decryption patterns in binary input data. Translated these patterns to Python and then C
  • Designed and implemented a Key Analyzer to read Excel input files, graph the given data, and perform operations to potentially reveal patterns in the data
  • Usability features include:
    • Automatically generated and named template python script to help test possible patterns
    • Auto-generated terminal window to test the script
    • Drag-n-drop GUI for file selection
  • By the end of the summer, we had solved enough patterns to meet the goal set for us

Pluralsight

Editor for hack.guides()

  • Currently working on Pluralsight's hack.guides() team to proofread open-source technical guides written for and by professionals around the world
  • Guides and tutorials covered a variety of CS topics. Check them out here!
  • Contributed to around 250 guides overall
  • Confident with Markdown formatting
  • Collaborated with the Pluralsight e-books team to categorize and catalog around 6,500 ebooks for the Pluralsight mobile app
  • Took on a part-time temporary role in September to work more closely with writers

Saijo Lab at Berkeley

Comp. Bio. Undergraduate Research Assistant

  • Working as a lab assistant at a neuroimmunology lab in UC Berkeley's Molecular and Cell Biology (MCB) department
  • Dr. Saijo's Lab researches microglial defense in the brain, especially immune response in glial cells
  • Responsibilities as an undergrad comp. bio. assistant:
    • Using HOMER libraries to run gene analysis scripts
    • Performing hierarchical clustering using Java TreeView and Cluster 3.0
    • Using HOMER's integration with R to find differential gene expression
    • Running STAR alignment jobs on Berkeley's High Performance Computing (HPC) cluster
    • Creating time-saving bash scripts to automate labor-intensive data-transfer tasks done using HPC

Education

University of California, Berkeley

Graduating May 2020

Electrical Engineering and Computer Science (EECS - B.S.)

Relevant coursework:

The Harker School

Graduated May 2016

High School

Projects

Turtlebot Tag

  • Interactive game of tag using Turtlebot robots.
  • Designed a pursuit-evasion game that integrated computer vision and robot movement principles.
  • Built using ROS, rospy, Turtlebots, and Intel Realsense camera.
View Project Website
*Need UCB Login

Predicting California's Storage Capacity Needs

  • Predicted where, when, and how much energy storage will be installed by 2024.
  • Energy storage is critical for state-wide wind and solar adoption since it limits curtailment and stabilizes the grid.
  • Mined features using CAISO interconnection queue data and geopolitical, socioeconomic data from the EIA and CSAC.
  • Built a f1-weighted voting system to predict which queued production projects would succeed.
  • Incorporated under-sampling techniques from imblearn to combat intrinsic dataset imbalance.
  • Full score on our project poster!
View Project Poster

Echoless

  • Goal was to visualize biases in a Twitter feed.
  • Performed sentiment analysis on Tweets to find keywords that users frequently tweeted with significant positive or negative sentiment.
  • Produced a Keywords vs. Sentiment Graph to highlight which keywords with most sentiment.
  • Plotted a user's tweet bias on a political spectrum from 0 to 24.
  • Used Python, Flask, JavaScript (AJAX), D3.js, HTML, CSS
  • Used Google's ML API for NLU (Keywords Graph) and Watson's API for NLC (Political Spectrum -- Supervised Training)
View Project On Github

Bear Maps

  • Created an online map of Berkeley using a collection of 16000+ map images.
  • Used a QuadTree data structure to raster the image query sent in by the user.
  • Worked with OpenStreetMap interface to pinpoint locations in the map and perform A* shortest-path routing.
  • Built an autocomplete feature using Tries to help suggest locations on the map as the user typed.
*Coursework code available by request!

SQL

  • Built a SQL system using Java as the back end.
  • Handled various SQL query features, including joins and where clauses.
*Coursework code available by request!

Image Blurring Using Python

  • Created a Moving Average Filter to simplify the input image into distinct colors.
  • Built this filter using queues in Python to handle RGB data.
  • Ultimately aimed to automate cropping by using k-means clustering to find the most relevant parts of an image.
View Project On Github

Bill Splitter

  • Used Python 2's OpenCV module to detect hand-drawn figures on the receipt image and to split the receipt into line items.
  • By the end, we were able to accurately detect a range of polygons and circles.
  • Final goal of this project was to automate bill splitting by processing an image of a receipt. A marked receipt would be submitted online and split into subpayments. Then, a payment request would be sent to group members via SplitWise.
View Project On Github

Organizations

ANova CS Mentors

Skills

Contact