Neeraj Kumar

Built a real-time hand-gesture control system using MediaPipe for detection and OpenCV for image processing. Implemented a gesture classification approach using a CNN model and exposed control through a Tkinter GUI. Used Python tooling to drive cursor interactions based on the recognized gestures.

Developed an ML pipeline using EDA and model evaluation metrics such as accuracy, precision, and recall. Trained classifiers with Random Forest and Logistic Regression and validated performance using cross-validation. Delivered a Tkinter GUI that provides real-time risk prediction based on the trained models.

Created an AI-driven healthcare platform feature that provides personalized treatment recommendations using Gemini with prompt engineering. Implemented backend APIs with secure authentication and a PostgreSQL data layer, then tested endpoints using Postman. Integrated a React UI styled with Tailwind CSS to support the overall user experience.

Built a Python automation pipeline that tests 950+ stream URLs, detects dead links, and updates playlists by pulling replacements from upstream sources. Scheduled the automation using GitHub Actions to run without manual intervention. Validated stream health with FFmpeg tooling and connected a Flutter/Dart Android TV app to load playlists from GitHub.

Investigated Python modules and SQL issues end to end by identifying root causes and implementing fixes to improve reliability. Wrote and optimized MySQL/PostgreSQL queries on large datasets and validated data accuracy. Built and tested REST APIs for ML-backed features using Postman, automated repetitive ETL-style validations with Python, and deployed containerized services on AWS (EC2, S3, Lambda). Developed AI-assisted document Q&A/RAG workflows and documented resolution procedures for the team.

Performed data preprocessing, EDA, and model evaluation on real-world datasets using Python. Used SQL to support data validation and documented findings in a shared repository. Built and assessed models with Pandas and scikit-learn, focusing on analysis quality and repeatable reporting.