26 ML projects, one repo, zero portfolio anxiety

What it does This repository bundles 26 machine learning projects across six domains—healthcare AI, computer vision, classical ML, NLP, time series, and geospatial—into a single portfolio. Roughly half are Jupyter notebooks; five are deployed web apps and three are standalone GUI applications. Think of it as a structured walk through the greatest hits of applied ML: tumor detection from MRI, driver drowsiness alerts, loan default prediction, and a symptom-to-diagnosis chatbot.

The interesting bit The author bothered to deploy. Most learning repos stop at model.fit(); this one wraps several projects in Flask and packages others as OpenCV-powered desktop apps. The README also tags every project by difficulty—beginner, intermediate, advanced—which saves you from accidentally opening a Kaggle MoA competition notebook when you wanted iris classification.

Key highlights

• 9 computer vision projects, from lane detection to real-time emoji overlays via webcam
• 6 healthcare projects including ECG arrhythmia classification and brain tumor detection
• 5 deployed web apps (Flask) and 3 GUI apps, not just notebooks
• Explicit difficulty ratings and tool callouts (PyTorch, scikit-learn, OpenCV, IBM Watson) per project
• Covers the boring-but-useful stuff: class imbalance (SMOTE), multi-label prediction, time series ensembling

Caveats

• The “~92% accuracy” claim for heart disease prediction is stated without validation details or confidence intervals
• Several projects rely on pre-trained models (Zhang et al. colorization, Caffe age/gender models) rather than training from scratch
• README truncates mid-sentence on time series and geospatial sections, so project counts there are unclear

Verdict Solid if you’re a student or self-taught developer assembling your first ML portfolio. Skip it if you need production-grade code or novel research—this is demonstrative work, not infrastructure.