Technologies: Python, scikit-learn, Pandas
•
Trained a classification model on [X sensor readings across Y fields] of soil-moisture and environmental data to predict on/off irrigation decisions per sprinkler, then serialized it with pickle and wired it into a standalone Python app (app.py) so it serves real-time predictions instead of staying a one-off notebook result.
Movie Recommendation System
Technologies: Python, Pandas, Scikit-learn
•
Built a content-based recommendation engine that vectorizes movie metadata (genre, cast, keywords/overview) and ranks titles by similarity score to return the top-N closest matches for a given movie, packaged as reusable preprocessing and scoring steps and validated with a recorded end-to-end demo.
Plant Disease Detection System
Technologies: Python, TensorFlow, OpenCV, CNN
•
Built and trained a CNN-based image classifier on a labeled leaf-image dataset, using augmentation (rotation, zoom, flips) and transfer learning to raise validation accuracy from [baseline X%] to [final Y%], which let the model correctly flag [Z] disease classes early enough to recommend timely, targeted treatment instead of a farmer having to wait for visible crop-wide damage.