< Back

All Projects

Analysis of TFSI Crystallographic Structures

Developed data engineering pipelines to efficiently process and visualize atomic properties of 1500+ structures. Reduced analysis time from 6 months to 5 minutes.

PythonMatplotlibPandasNumPyPlotly

Optimized RFE Feature Selection with LogRes

Developed a custom Logistic Regression model for binary classification on heart attack risk data, optimizing feature selection while maintaining 94% accuracy.

PythonNumPyStandardScaler

MathVenture

A speed math platform that dynamically generates math problems and analytics using a custom math engine.

Next.jsSupabaseTailwind CSSPostgreSQL

Email Classifier

Performed sentiment analysis on 30,000 emails using TfIDF vectorization and developed a classifier with 97.8% accuracy.

PythonsimpleGmailSklearnMatplotlibNumPyPandas

Tree

Developed a strategy game that promotes sustainability in an entertaining medium. Implemented Unity ScriptableObjects to create different tree templates and designed a biodiversity mechanic essential to winning the game.

Unity APIC#ScriptableObjects

Zer0

Received 160+ installs on Google Play across 15 countries and was featured on local gaming blogs. Developed a first-person 3D Android parkour game with physics-based movements like wall running, grappling, and jetpacks.

UnityBlenderC#GIMP

EirAI

Theme Track 2nd Place - HackKU 2025

A service designed to reduce administrative bloat for healthcare professionals and medical coders. EirAI converts patient-doctor conversations into structured SOAP notes using speech-to-text, identifies relevant ICD codes, and cross-references a dataset of prior insurance approvals to suggest the most approvable treatment paths. The system indexes ~73,000 ICD codes and simulates real-world conditions using a dataset with over 1.5 million rows.

PythonWhisperGPT-4Google GeminiPostgreSQLNext.jsFramer-MotionTailwind CSS

Decentralized Artificial Intelligence Marketplace (DAIM)

Autonomys AI Track Winner - Midwest Blockathon 2025

A platform where developers can share open-source AI models without losing ownership or fearing platform obsolescence. DAIM provides structured access to AI models with performance metrics for seamless integration and customization. Contributors can securely publish and attribute their work while leveraging generative AI for enhanced model insights.

Pinata IPFSNext.jsAutonomy APITypeScriptGoogle Gemini APITailwind CSSEthers.js

Development of Cheminformatics Pipelines to analyze Crown Ethers

Currently developing novel analysis techniques using graph theory to analyze the properties of crown ethers. This will be used to predict the binding affinity of crown ethers to various metals.

PythonMatplotlibPandasNumPyPlotly