In elementary school, I was intrigued by how the internet works. At the 4th grade book fair, I found an HTML book and built a website explaining my science project. That was my first coding project, and I was excited to see how quickly I could build a product that anyone could use. I wanted to code more.
In middle school, I learned Python and Java through many projects.​ I then decided to join The Harker School because of its strong focus on STEM and because it offered many advanced Computer Science courses. With my continued interest, I was also selected to be one of the developers in the HarkerDev team to build and deploy web applications for students and teachers.
Additionally, I have been intrigued by AI since my first AI camp in my first year. While volunteering at the South Asian Heart Center, I thought of ideas to help them use AI. It has been exciting to continue learning through a variety of projects to impact my community, and I am looking forward to pursuing this further through my college and career journey.
AI Projects
Heart Disease Risk Estimator Chatbot
Created a chatbot to inform users of their heart disease risk factors based on age, gender, and cholesterol level. The chatbot uses Open AI’s GPT 3.5 engine in the backend and streamlit.io as UI to provide an easy-to-use interactive chat interface for collecting relevant information from the user. It then uses the data and calculates the user’s heart disease risk probability using an online calculator based on the Framingham Risk Score!
Grade: Rising 11th
Programming Language: Python, Streamlit.io, Open AI GPT 3.5/4.0 LLM, Prompt Engineering
Code: https://github.com/sbhorkar/heartRiskChatBot
Video demo: https://drive.google.com/file/d/1sQtt0MSfROHdhXRyIVVJtn3db4eTXCJ-/view?usp=sharing
Web application:
Mental Health Chatbot
As a part of the Inspirit AI winter camp, we built a mental health chatbot using NLP. We used RNNs and LSTMs and even built our custom transformer. We used a pre-made dataset to create this chatbot as well. Google Slides Presentation here.
Grade: 10
Programming Language: Python, TensorFlow
Code: https://colab.research.google.com/drive/1KVSAmYDkSKKAkattVsUPM61rN925MXzq?usp=sharing
​
Applying Machine learning to create a heart disease risk estimator
As part of research with Dr. Goldzstien, a Professor at Georgia Tech University, I trained an ML model that can estimate the risk of cardiovascular disease using publicly available data from Kaggle about 320,000 patients. The model used about 18 features and the TensorFlow Sequential Neural Network algorithm to train the model and achieved a 77% accuracy. The purpose of the model is to replace calculators, such as the one I used in my chatbot, that use far fewer factors and a linear regression model. I have submitted this research project for publication in the Journal of Student Research.
Grade: Rising 11th
Programming Language: Python, TensorFlow
Code: https://colab.research.google.com/drive/1KVSAmYDkSKKAkattVsUPM61rN925MXzq?usp=sharing
Programming Projects
PASCAL Compiler
Building a compiler for a subset of PASCAL. So far, I have built the first three phases: syntax analysis, lexical analysis, and semantic analysis.
Grade: 11
Programming Language: Java
HarkerDev
As a member of my school's student developer team, I built a feature to simplify updating the lunch menu on an internal portal that students use to check what's for lunch every school day. Before this feature, an administrator would need to update this content manually, so this feature saved 10 mins a day for the admin. The feature is deployed to production and used by 900 students and teachers.
Grade: 10, 11
Programming Language: Vue, Javascript
Code: Lunch Schedule Editor Code
​
​
Nectar (previously called Cancer Connect)
Nectar is a social media app designed specifically for the cancer community. The idea for this app, which is currently in alpha, was developed by me and three other high schoolers, including one who has battled cancer for four years. Based on this idea, we received a grant that engaged a third party to develop the app. Once on the platform, users can connect with like-minded individuals (i.e., cancer parents can connect with others). They can set up discussion threads, direct messaging, and group chats.
Grade: 10, 11, 12
Technology Used: Figma
Google Slides Pitch linked here
App Demo linked here
PetBit 1
An AdaFruit-based product to track pet's steps and provide nighttime visibility and temperature control for pet owners.
Grade: 7, ProjectCSGirls Honorable Mention
Technology Used: AdaFruit, Python
Code: https://github.com/petbit-team/PetBit
Video:https://drive.google.com/file/d/14E7JWo6X58s3omQCGTI1tkpfoZGwztXh/view?usp=sharing
Technical Report: PetBit 1
Tetris Game
Enjoyed building the Tetris game as I was learning Data Structures. It has features like score and rotating capabilities.
Grade: 10
Programming Language: Java
PetBit 2
I evolved PetBit 1 into a more convenient solution that used Puck.JS and Javascript. I also added location tracking using MFi. There is no pet device in the market today that includes activity tracking, location tracking, and training aids; PetBit 2 will have all three. After a successful prototyping in the summer before junior year, I plan to launch a Kickstarter campaign the following summer.
Grade: 10,11,12
Programming Language: Javascript, HTML
CarBurn Footprint
Carburn Footprint is an application that lets users know how much carbon their car will emit based on make, model, trim, and miles driven annually. It can also be used as a car purchase decision tool, as users can decide what car to buy based on how much carbon it releases into the atmosphere. The data is taken from fueleconomy.gov.
Grade: 8, ProjectCSGirls Finalist
Programming Language: Java, JavaFX for UI, MySQL database, JDBC
Code: github.com/sbhorkar/CarBurnFootprint.
Video: https://youtu.be/MqZG6DiNqvk
Technical Report: CarBurn Footprint
​
​