💳 Build and track credit risk models using MLflow for experiment management, feature engineering, and model selection in loan default prediction.

🔍 Predict customer churn with a machine learning system that identifies at-risk clients and recommends tailored retention strategies for better ROI.

📧 Detect spam emails with ease using machine learning and the Naive Bayes algorithm for fast, accurate results.

🩺 Predict diabetes risk using an end-to-end machine learning pipeline, featuring advanced models and techniques for superior accuracy in the Kaggle competition.

📊 Explore panel time-series forecasting techniques for sales using popular Python libraries like ARIMA, Prophet, and AutoTS in Jupyter notebooks.

❤️ Predict heart disease risk using classic machine learning techniques with this Jupyter notebook project, featuring data exploration and model building.

🩺 Predict pancreatic disease status using advanced machine learning techniques with a focus on reproducibility and accuracy in biomarker data analysis.

Real-time Web Dashboard for Optuna.

This repository contains solutions for Kaggle's Playground Series prediction competitions. It showcases a structured and analytical approach to machine learning, covering both regression and classification tasks.

Modular PyTorch framework: Pydantic schemas + Optuna optimization + resolution-aware architectures for vision research

Machine learning lead scoring system that classifies sales inquiries into Hot, Warm, Cold, Save for Later, or Reject using CatBoost and business-driven features.

MLB game outcome predictor trained on 2.1M Statcast pitches — XGBoost + Optuna achieving 0.629 ROC-AUC, outperforming Vegas baseline

This project implements a high-performance pipeline for aerodynamic shape optimization. It uses Bayesian Optimization to discover ideal NACA 4-digit airfoils across a flight envelope and trains a Random Forest Surrogate Model to provide instantaneous aerodynamic predictions.

Machine learning model to predict customer churn | Kaggle Competition | Score: 0.91532 | XGBoost, LightGBM, Feature Engineering

Deep reinforcement learning for quadruped locomotion in Genesis: comparative study of PPO, SAC, and TD3 on flat and irregular terrains. Unitree Go2, Stable-Baselines3, Optuna. CROS 2026.

Production-grade MLOps pipeline for aircraft engine failure prediction on NASA CMAPSS turbofan data. FastAPI + Streamlit + MLflow + Optuna. Deployed on Hugging Face & Streamlit Cloud.

Quant AI backend: FastAPI service for market data, feature engineering, model training, and explainability (SHAP).

Credit risk prediction pipeline with MLflow experiment tracking. XGBoost, LightGBM, Optuna tuning, SHAP analysis

A project to deploy an online app that predicts the win probability for each NBA game every day. Demonstrates end-to-end Machine Learning deployment.

CatBoost classifier for heart disease prediction trained on 630K records with Optuna 50-trial hyperparameter tuning, achieving ROC-AUC of 0.9562. FastAPI backend and Streamlit UI for single and batch predictions. Training runs on Modal T4 GPU when deployed and falls back to CPU locally.