Hi, my name is Harsha Karanth


Hello, my name is Harsha Karanth, and I am a current masters student at the Georgia Institute of Technology focusing on machine learning and application development. I pursued the computer science field because I felt it was the best way to impact people’s daily lives. I strive to create applications that people use everyday and have a positive impact on the world. I grew my technical skills by pursuing side projects and working at Google and Apple as an Intern. I also joined the Polo Club of Data Science this August to research machine learning capabilities and display its impactful results to the general public.

Fun Fact: I am an Eagle Scout!

Work Experience

Machine Learning TA

August 2023 - Current

Hosting Office Hours to teach machine learning concepts and guide students on their homework/projects.

Creating and grading homeworks, projects, and quizzes that enhance student’s machine learning knowledge.

Apple SWE intern

May 2023 - August 2023

Created Machine Learning/Computer Vision models using Python to automate testing for a Video Streaming Team.

Google SWE intern

May 2022 - August 2022

Recorded Android camera metadata and started the replaying process of failed tests to save developers time.

Created an Android App using Java for Google Pixel developers so they can efficently record camera metadata.

Google STEP intern

May 2022 - August 2022

Developed two IOS test apps using Objective-C that incorporates a Driver SDK to display ridesharing and delivery driver's workflow.

Utilized the IOS URL Loading System to send HTTP requests and serialization classes to parse JSON.

My Current Projects

Cardiac AR App

August 2023 - TBD

Building AR features using Swift and the RealityKit library to create an IOS app that allows doctors to view, rotate, and see the cross-section of a 3D heart.

Using quaternion math to calculate the correct angle to rotate and cut the heart and using custom metal shaders to display the changes.

Creating a multi-view AR model in which multiple people on different devices can change and use the same AR model.

Diffusion App

August 2023 - TBD

Creating an IOS app using Swift with diffusion models that generate AI portraits and performs all computation on-device.

Initially, I am exploring open source diffusion apps, benchmarking performance, and optimizing diffusion models

Github coming soon!

SEC 8K Filings Stock Prediction

March 2023 - TBD

Analyzing the quality of financial agreements that are listed in company's 8K filings in section 1.01 and 2.03 and then discovering trends that can used for stock prediction using Yahoo Finance and ThinkOrSwim.

Created a Python script using the Selenium and Beautiful Soup libraries to web scrape sections of 8K filings

Preprocessed text data by removing stop words, performing lemmatization, and then tokenizing words

Vectorized the data using TF-IDF, trained Naive Bayes and SVM models on this data, predicted the quality of financial and material agreements on new companies, and analyzed stock movements during agreement periods

Egyptian Rat Screw (ERS) Mobile App

January 2020 - TBD

Used Unity and C# to create a multiplayer card game in which users can play against other people to increase their rank, create a custom lobby to play with friends, and play against the computer.

Leveraged the Photon Server to host the games and Google Saved Games/iCloud to store player data.

Implemented RPC calls to create the multiplayer feature and bitwise operators to reduce data storage by 85%

My Completed Projects

Monet-Style Image to Image Translation (Team Project)

May 2023

Created a Deep Conventional Generative Adversarial Network (DCGAN) and Cycle Generative Adversarial Network (CycleGAN) using Python to convert pictures of buildings and landscapes into Monet-styled images

Implemented discriminators and generators for both DCGAN and CycleGAN using Tensorflow layers

Used MiFID error to measure the difference between the input and output images of the two models and experimented with image augmentation and changing hyperparameters such as learning rate to reduce error

Warehouse Manager

July 2022

Utilized Angular and Bootstrap to create an application that can track goods in the warehouse by allowing workers to upload information about a new shipment and see all the current shipments in the warehouse.

Employed services to share information from the form to the component that displays the shipments and used Bootstrap Accordion to allow workers to see a summarized and expanded version of the orders

Used Express.js to create a RESTful API and MongoDB to store data and relationships

Empathy Bytes IOS Mobile App (Team Leader)

Spring 2021 - May 2022

Worked in and led a team to bring to light influential communities at Georgia Tech by creating an IOS mobile app that displays each community's culture by using Virtual and Augmented Reality.

Used Blender and RealityKit to create models of the Hive's machines and products and create an interactive game with the models so that Georgia Tech students can experience the exciting tools the Hive can offer.

Leveraged Matterport and Google Cardboard to create 360 virtual tours of the Hive makerspace.

Experienced the lack of documentation with RealityKit and SceneKit and created documentation for new members.

Contact Me

I'd love to get in contact with you, so please feel free to reach me on any of these platforms.

Template from Rosacodes