Skip to main content
kush@ml-engineer: ~/portfolio
ACCURACY
0.00%
LOSS
100.000
EPOCHS
0/100
TRAININGGPU: RTX 4090RAM: 64GB
FRAMEWORK: PyTorch
INITIALIZING PORTFOLIO EXPERIENCE
KUSH RANK
KUSH RANK

SCROLL

About Me

Brief

I'm Kush, a data science and ML engineer-in-progress, turning messy real-world datasets into clean insights, practical models, and small "Jarvis-style" tools — from flight prices to food delivery — and shipping them as web apps people can actually use.

0+
Projects Completed
0
ML Models Deployed
0
Hackathon Finalists
🎓

Education

New Jersey Institute of Technology

Bachelors of Computer Science • 27'

🎯

Focus Areas

Machine Learning • GenAI

End-to-end ML pipelines

🚀

Currently Learning

MLOps • PySpark • Deep Learning

Scaling ML to production

💼

Status

Open to Opportunities

Internships • Full-time roles

THE ARSENALTHE ARSENAL

💻

Languages

PythonJavaScriptTypeScriptJava
Click to flip
💻

Languages

Core Proficiency

Click to flip back
🤖

ML & AI

Scikit-learnPyTorchTensorFlowKerasMLFlowXGBoost
Click to flip
🤖

ML & AI

5+ Projects • Advanced

Click to flip back
🧠

GenAI & LLMs

LangChainTransformersAgentic AIGenAI
Click to flip
🧠

GenAI & LLMs

Cutting Edge

Click to flip back
📊

Data Science

PandasNumPyEDAMatplotlibPlotly
Click to flip
📊

Data Science

Expert Level

Click to flip back
🌐

Web & Full Stack

Next.jsReact.jsNode.jsTailwindCSSThree.jsFlaskStreamlit
Click to flip
🌐

Web & Full Stack

4+ Projects

Click to flip back
🗄️

Databases

MySQLMongoDB
Click to flip
🗄️

Databases

Production Ready

Click to flip back
🚀

DevOps & Tools

DockerGit + GitHubCI/CDn8nJupyterVS CodeConda/venv
Click to flip
🚀

DevOps & Tools

Daily Workflow

Click to flip back
🔥

Currently Learning

PySparkMLOpsDeep LearningLightGBMCatBoostCNNsRNNs
Click to flip
🔥

Currently Learning

In Progress

Click to flip back

PROJECTS

01
PROJECT #01

UniWallet - Smart Campus Wallet

CHALLENGE

College students juggle 5+ apps for banking, meal plans, budgeting, and payments, leading to fragmented financial management and missed spending insights.

APPROACH

  • Integrated Plaid API for secure bank account linking in sandbox mode
  • Built AI Financial Coach (FinBot) using Ollama Cloud for personalized advice
  • Implemented smart transaction categorization across dining, transport, books, events
  • Created unified wallet system consolidating campus wallets and meal plans

IMPACT

Developed in 48 hours at HackFest 2025 @ Rutgers Newark, delivering a fully functional AI-powered financial platform.

Top 5 Finalist

TECH STACK

Next.js 16React 19.NET 8MySQLOllama AIPlaid APITypeScript
02
PROJECT #02

Student Performance Predictor

CHALLENGE

Educators need to identify students requiring additional support by predicting academic outcomes from demographic and performance data.

APPROACH

  • Evaluated 8 regression algorithms (Linear, Ridge, Lasso, Random Forest, XGBoost, CatBoost)
  • Implemented separate preprocessing pipelines for numerical and categorical features
  • Built responsive Flask web interface for real-time predictions
  • Deployed production-ready application on Render.com

IMPACT

Analyzed 1,000 student records with automated model selection using 5-fold cross-validation.

Best Model Selected
R² optimization
Production Deployed

TECH STACK

PythonFlaskScikit-learnXGBoostCatBoostPandasRender
03
PROJECT #03

Jarvis - Personal AI Assistant

CHALLENGE

Daily task automation through voice commands requires seamless integration of speech recognition, web APIs, and system utilities.

APPROACH

  • Implemented voice interaction with speech-to-text and text-to-speech conversion
  • Integrated OpenWeather and WolframAlpha APIs for real-time data
  • Built modular architecture supporting custom commands and plugins
  • Created system automation for app launching, screenshots, and file management

IMPACT

Voice-powered AI assistant with 5+ capability categories and extensible plugin system.

Fully Functional

TECH STACK

PythonSpeech RecognitionOpenWeather APIWolframAlpha APIText-to-Speech
04
PROJECT #04

Flight Price Prediction

CHALLENGE

Travelers struggle with unpredictable flight pricing patterns and lack tools for informed booking decisions.

APPROACH

  • Cleaned and analyzed 10,000+ flight records from Excel datasets
  • Engineered 15+ time-based and route-based features
  • Performed extensive EDA on journey dates, airlines, duration, and pricing correlations
  • Implemented feature encoding and scaling for ML readiness

IMPACT

Comprehensive data preparation pipeline ready for model training and deployment.

EDA Complete

TECH STACK

PythonPandasNumPyMatplotlibSeabornScikit-learn
05
PROJECT #05

Sales EDA & Purchase Prediction

CHALLENGE

Understanding sales patterns and building accurate forecasting models requires comprehensive data exploration and feature engineering.

APPROACH

  • Performed extensive EDA with distribution, trend, and correlation analysis
  • Handled missing values and corrected data types systematically
  • Implemented and compared multiple regression models (Linear, Random Forest, XGBoost)
  • Evaluated models using RMSE and R² metrics with prediction visualization

IMPACT

End-to-end data science workflow from raw data to model evaluation.

Multi-model comparison
RMSE + R² metrics
Complete Pipeline

TECH STACK

PythonPandasNumPyScikit-learnXGBoostMatplotlibSeaborn

In Development

More Projects Incoming

Building the future, one commit at a time

READY TO BUILD
SOMETHING TOGETHER?

Let's turn data into impact

Designed & Built by Kush Rank

Tech Stack:

Next.js 14 • React 18 • TypeScript • Tailwind CSS • GSAP • Lenis

© 2025 | Crafted with precision & Coffee ☕

Click to chat with my Personal Assistant
🤖 Hi, I'm JARVIS!
Kush's Personal Assistant
Ask me anything about Kush