Extended Project Portfolio

Built a screening-oriented ML system for early diabetes risk estimation, with emphasis on recall-focused evaluation for high-sensitivity healthcare use cases.

Designed the workflow with preprocessing, model inference, and a simple application layer to connect ML outputs to a usable screening interface rather than a notebook-only experiment.

Focused evaluation choices on realistic screening behavior rather than accuracy alone, improving the project’s relevance for practical healthcare classification scenarios.

Deployed the solution through FastAPI and a lightweight Streamlit interface, making the workflow easier to interact with and demonstrate end-to-end.

Developed a predictive system for identifying phishing websites using machine learning classification, applying feature-based modeling to a real cybersecurity detection use case.

This project demonstrates practical ML application in security-focused pipelines, where detection quality and classification reliability matter more than only experimentation.

Built a forecasting workflow for airline passenger demand using statistical and machine learning approaches, with outputs delivered through an interactive Streamlit interface.

Compared ARIMA, ETS, and ML-based forecasting strategies to understand tradeoffs in time-series modeling and practical demand estimation.

Created a FastAPI-based backend paired with a Streamlit frontend to simulate a lightweight e-commerce workflow with API-driven interactions and structured service logic.

This project highlights backend fundamentals, endpoint design, and practical API-to-UI integration using a clean Python service layer.

Built a backend weather application with API integration, validation, persistence, CRUD operations, and data export functionality in a structured FastAPI service.

The project demonstrates backend API design, real-time data handling, and service-oriented thinking for practical application workflows.

Explored sentiment classification on movie reviews using recurrent neural networks, focusing on how sequence models behave on real-world text data.

This project helped build understanding of text preprocessing, sequence modeling limitations, and practical NLP experimentation on benchmark-style sentiment data.

Developed an ANN-based churn prediction workflow with preprocessing, modeling, and inference-style structure for customer retention analysis.

The project demonstrates supervised learning for tabular business problems and connects predictive outputs to customer-focused decision support scenarios.

Practiced cloud provisioning concepts by defining AWS EC2 infrastructure through Terraform, reinforcing infrastructure-as-code and repeatable environment setup.

This project reflects growing deployment and infrastructure awareness alongside application development, especially around cloud-ready engineering workflows.

Built reporting and dashboard-style analytics to transform business data into more decision-friendly views, focusing on clarity, insight generation, and stakeholder-facing presentation.

This project demonstrates practical analytics communication and reporting design, not just raw data modeling.

Performed exploratory data analysis on country-level data to identify distributions, patterns, and relationships through Python-based analytics workflows.

This project reflects core analytical thinking and data storytelling using tabular datasets, with emphasis on interpretation rather than only model building.