A portfolio platform showcasing business analytics, machine learning, predictive modeling, and executive recommendations — every project shipped as a fully interactive dashboard.
Turn 1,898 food-delivery orders into customer, operational, and revenue insights with executive-ready recommendations.
Predict CNC machine failures from sensor data using an in-browser decision tree, with feature importance, evaluation, and a live failure-risk simulator.
Predicting loan acceptance using supervised learning classification. Helps AllLife Bank convert depositors into loan customers with targeted, efficient marketing.
Predict US visa certification outcomes for 25,480 OFLC cases using Decision Trees, Bagging, Random Forest, AdaBoost, and Gradient Boosting — with resampling and hyperparameter tuning.
Predict wind turbine generator failures from 40 ciphered sensor signals with a tuned neural network, cutting maintenance cost by 64.9% on unseen data.
A RAG assistant over the 4,114-page Merck Manual that answers clinical questions with page citations — compared against a plain LLM and prompt engineering, and scored by an LLM judge.
A LangGraph multi-agent system that triages delivery exceptions, picks a playbook-grounded resolution, escalates by policy and writes tier-appropriate customer messages — 10/10 test cases correct.
Predict loan default for 39,933 NBFC borrowers with tuned LightGBM — 89.6% cross-validated accuracy against a 76.3% no-model baseline.