Create README for February 2026 AI/ML experiments

Added README for February 2026 AI/ML local experiments with project details, learning goals, and resources.

Author: SnakeEye-sudo

2026-02-local-ml/README.md

@@ -0,0 +1,81 @@
+# 🤖 February 2026: AI/ML Local Experiments
+
+## 📌 Theme Overview
+
+**Focus**: Local Models, Datasets, Prediction - Building ML projects that run entirely in the browser
+
+**Philosophy**: No black-box APIs, only transparent math and reproducible experiments. Every model runs locally, every dataset is included, and every prediction is explainable.
+
+## 🎯 Learning Goals
+
+- Understand ML fundamentals: features, labels, loss functions, optimization
+- Implement regression and classification from scratch
+- Work with datasets: preprocessing, normalization, train/test splits
+- Build inference pipelines that run in the browser
+- Create explainable AI demos with visual feedback
+
+## 📂 Projects
+
+### Project 1: Local ML Playground
+**Type**: Exploration
+**Tech**: Pure JavaScript, Chart.js
+**Description**: Interactive regression sandbox that trains models in the browser. Users can adjust features, see training progress, and understand how gradient descent works.
+
+**Features**:
+- Live regression training with visualized loss curves
+- Interactive parameter controls (learning rate, iterations)
+- Real-time prediction updates
+- "Show Math" mode explaining coefficients and calculations
+
+### Project 2: On-Device Classifier
+**Type**: Showcase
+**Tech**: TensorFlow.js / Custom JS implementation
+**Description**: Lightweight classification demo using local model files. No external APIs, no server-side processing.
+
+**Features**:
+- Pre-trained model stored in repo (quantized for speed)
+- Text or image classification
+- Instant on-device inference
+- Model architecture visualization
+
+## 🚀 What Makes This Month Different
+
+- **Zero External Dependencies**: All ML logic runs locally
+- **Educational First**: Focus on understanding over performance
+- **Reproducible**: Datasets and models included in repo
+- **Browser-Native**: No Python, no backends, pure web tech
+
+## 📚 Key Concepts Covered
+
+- **Supervised Learning**: Training with labeled data
+- **Loss Functions**: MSE, Cross-Entropy
+- **Optimization**: Gradient Descent variants
+- **Preprocessing**: Feature scaling, normalization
+- **Evaluation**: Train/test splits, metrics (MAE, accuracy)
+- **Inference**: Forward pass, prediction pipeline
+
+## 🔗 Resources & Inspiration
+
+- Linear regression math: [Khan Academy](https://www.khanacademy.org/math/statistics-probability/describing-relationships-quantitative-data/introduction-to-trend-lines/a/linear-regression-review)
+- TensorFlow.js docs: [tensorflow.org/js](https://www.tensorflow.org/js)
+- Interactive ML visualizations: [Distill.pub](https://distill.pub/)
+
+## 📊 Project Status
+
+- [x] February folder structure created
+- [ ] Project 1: Local ML Playground
+  - [ ] Dataset selection and preprocessing
+  - [ ] Regression implementation
+  - [ ] UI with interactive controls
+  - [ ] Loss curve visualization
+- [ ] Project 2: On-Device Classifier
+  - [ ] Model selection and preparation
+  - [ ] Inference pipeline
+  - [ ] Demo UI
+  - [ ] Documentation
+
+---
+
+**Month**: February 2026
+**Status**: In Progress
+**Next**: Build both projects and update root README