Esha Karlekar

University of California, Berkeley

Electrical Engineering and Computer Sciences, B.S.

About

Hi! I'm a student at UC Berkeley majoring in Electrical Engineering and Computer Sciences (EECS) and minoring in Public Policy. I am primarily interested in artificial intelligence and the way it can be used to solve interesting problems in different fields of study, particularly in application to social good, education, and healthcare. However, I have experience and interests in other fields of computer science, including full-stack, data science, and general software engineering.

I recently finished my internship at Netflix this summer for their Trust & Safety team, where I worked on leveraging ML to pinpoint anomalies in time series data of authentication requests, preventing account takeover & fraud. I recently also took a semester off of school to do an internship at SpaceX, where I am doing ML/SWE work for their Special Programs GNC (Guidance, Navigation, and Control) Team. I have also previously interned/worked part-time at MITRE, working on generative AI models (specifically large language models) in conjunction with computer vision models in application to government record and satellite data analysis.

Exploring the intersection of business and tech, I am a part of Valley Consulting Group, Berkeley’s premiere tech-consulting group, working with Fortune 500’s and industry-leading tech companies. Most recently, I completed a project consulting for Airbnb and Puma, advising them on their artificial intelligence strategy in application to identity verification and their marketing scheme, respectively.

I've completed machine learning and full-stack research internships at the Cardiac Vision Laboratory at UCSF, the Howard Hughes Medical Institute Janelia Research Campus, the US Naval Research Laboratory, and George Mason University. In my free time, I try to give back to the community, volunteering at the local soup kitchen in Berkeley. I also recently volunteered as a Data Science Fellow at Bluebonnet Data, working on a project to consolidate voter records for a nonprofit organization working to make higher education free. I love to upcycle clothing, bake, watch TV, and go on long drives.

Personal Projects

Internship Projects Available for Public View

    HHMI Janelia Research Campus | 2020-2022

    Designed, implemented, and tested a machine learning system for protein identification in electron microscopy images, entitled 'brEMbow'. Rendered marker proteins using computer vision techniques and created training pipeline for convolutional neural network (specifically U-net). Additionally, worked on data analysis in Python for paper to be published in Nature in collaboration with researchers from the University College London, specifically key metrics (distance, density, vesicle type).
    BrEMbow Writeup
    BrEmbow Github
    Secondary Paper Preprint

    George Mason University | 2021

    Built a mobile app entitled "SnowSource" that parses snow and rainfall data in the U.S. to give users information about water supply and farmland viability. Worked on server-side and backend in JavaScript, Python, and Flask.
    Publication in GMU's JSSR Journal
    Read more and download app

DALÍ: Art at Your Fingertips | 2023

Messaging Sticker Extension Utilizing Generative AI Models for Image Generation

    Currently working on an app that serves as a keyboard extension app, allowing users to access the extension in their keyboard on the iMessage app, input a prompt, and generate an image from that prompt to use as a sticker in their conversation. The image generation utilizes OpenAI's DALLE image generation software. The app was developed primarily using Swift.

Github
Capstone Presentation

Blinks to Speech | 2021-22

Using EEGs to Detect Eye Blinks for Morse Code Communication

    As a part of a senior year-long research project, we implemented a system to translate eye blinks detected from electroencephalograms (EEGs) to Morse code to English text-to-speech. We conducted stream data processing of audio files from the EEG, translated blink data to English, and created a user interface. The system is meant to serve as a means of AAC (augmentative and alternative communication) for people with certain afflictions (locked-in syndrome, cerebral palsy, quadriplegia, etc.) where they are able to control their eye movements but are unable to move other parts of their body and therefore cannot communicate by conventional means. Since current means of communcation are extremely specialized to each individual and are therefore incredibly costly, our system therefore serves as a cheaper and more accessible means of communication. Utilized skills of accessible UI development and signal processing.

Github
Paper
Poster

Sentimeet: A Biometric Sentiment Analysis Extension for Google Meet | 2020

Hackathon Project Aimed at Improving Virtual Communication

    During the pandemic, as instruction and meetings shifted to a virtual setting, many of the cues that might give individuals a "read of the room", such as body language, became much harder to understand. The task of "reading the room" for teachers, students, people with autism or other neurological developmental disabilities that may affect the ability to read social cues, and people giving presentations became increasingly more complicated. SentiMeet serves as a solution to these issues. SentiMeet is a Google Chrome Extension that utilizes AWS's botocore 'rekognition' to analyze the faces of the group and automatically generates a report outputting the emotions of the group for up to 100 faces, formatted for readablility. For instance, given a screenshot of group of students, half of which are angry and half happy, the program will alert the user that the faces were 50% angry and 50% happy. Utilized Python, JavaScript, and Flask.

GitHub
DevPost

Hinder: Swipe Right for Change | 2020

Hackathon Project Aimed at Making Signing Petitions Easier and More Accessible

    Hinder is a website that allows users to easily scroll through petitions that support various movements and causes. Users can scroll through a vareiety of petitions and swipe right on the petitions they would like to sign. This system reduces the redundancy that other websites like change.org might encounter, so users do not have to constantly find and re-input their information to sign petitions. The system webscrapes change.org petitions and finds ones that would match the users interests based on past petitions signed. With the overwhelming amount of petitions there are, this allows users to access them easier and help make change, even if it is in a relatively small/simple way, such as signing a petition. Website utilizes Javascript, Python, HTML, and webscraping.

DevPost

Volunteering

Features

Senior Profile by UNC-CH and UNC School of Arts and Sciences

Recognized by UNC-CH Computer Science for Undergraduate Research

CRA Outstanding Undergraduate Researcher Award Runner-Up

Featured in Top Paper Picks in Sebastian Ruder's NLP Newsletter

Work Featured in Towards Data Science and Democratizing Artificial Intelligence Research

Featured and Interviewed by UNC Chapel Hill Admissions

Featured on Front Page of Daily Tar Heel in Women in Science Article

Featured as Exemplary Researcher in Endeavors Magazine

Honors

  • CRA Outstanding Undergraduate Researcher Award Runner-Up
  • Neo Scholar – Mentorship Community for Entrepreneurship
  • Distinguished Scholar - Chancellor's Science Scholarship Program
  • Ernest H. Abernethy Prize for Student Publication – Chancellor’s Award
  • Phi Beta Kappa Honor Society
  • EMNLP 2018 Student Scholarship
  • Moogfest Young Engineers Scholarship
  • Grace Hopper 2018 UNC Chapel Hill Scholarship
  • 1st Place Math and Computer Science Poster - National Sigma Xi Conference
  • STEM Diversity Scholarship – Full Scholarship, Academic Merit
  • Chancellor’s Science Scholars – 10k/yr Scholarship, Academic Merit
  • Rewriting the Code, Women in Computer Science Fellow
  • Dean’s List for All Semesters
  • National Merit Scholarship Finalist
  • National AP Scholar Award
  • Best Design for Mobile Application – HackTJ Hackathon

Elements

Text

This is bold and this is strong. This is italic and this is emphasized. This is superscript text and this is subscript text. This is underlined and this is code: for (;;) { ... }. Finally, this is a link.

Heading Level 2

Heading Level 3

Heading Level 4

Heading Level 5
Heading Level 6
Blockquote
Fringilla nisl. Donec accumsan interdum nisi, quis tincidunt felis sagittis eget tempus euismod. Vestibulum ante ipsum primis in faucibus vestibulum. Blandit adipiscing eu felis iaculis volutpat ac adipiscing accumsan faucibus. Vestibulum ante ipsum primis in faucibus lorem ipsum dolor sit amet nullam adipiscing eu felis.
Preformatted
i = 0;

while (!deck.isInOrder()) {
    print 'Iteration ' + i;
    deck.shuffle();
    i++;
}

print 'It took ' + i + ' iterations to sort the deck.';

Lists

Unordered
  • Dolor pulvinar etiam.
  • Sagittis adipiscing.
  • Felis enim feugiat.
Alternate
  • Dolor pulvinar etiam.
  • Sagittis adipiscing.
  • Felis enim feugiat.
Ordered
  1. Dolor pulvinar etiam.
  2. Etiam vel felis viverra.
  3. Felis enim feugiat.
  4. Dolor pulvinar etiam.
  5. Etiam vel felis lorem.
  6. Felis enim et feugiat.
Icons
Actions

Table

Default
Name Description Price
Item One Ante turpis integer aliquet porttitor. 29.99
Item Two Vis ac commodo adipiscing arcu aliquet. 19.99
Item Three Morbi faucibus arcu accumsan lorem. 29.99
Item Four Vitae integer tempus condimentum. 19.99
Item Five Ante turpis integer aliquet porttitor. 29.99
100.00
Alternate
Name Description Price
Item One Ante turpis integer aliquet porttitor. 29.99
Item Two Vis ac commodo adipiscing arcu aliquet. 19.99
Item Three Morbi faucibus arcu accumsan lorem. 29.99
Item Four Vitae integer tempus condimentum. 19.99
Item Five Ante turpis integer aliquet porttitor. 29.99
100.00

Buttons

  • Disabled
  • Disabled

Form