An empirical comparison of deterministic vs. stochastic exploration in stationary Multi-Armed Bandits.

Real-time fraud detection — sliding-window velocity features, impossible travel, spend z-scores, explainable rule DSL with hot reload, online logistic model, review queue, HTTP scoring API, synthetic traffic with 4 fraud patterns. Live simulator. Prism Systems 09/12

A machine learning-based Rice Crop Yield Forecasting System that analyzes agricultural and environmental factors to predict rice yield and support data-driven farming decisions.

Training of gradient-network magnetic models

Practical, reproducible training and fine-tuning guides for embeddings, rerankers, LLMs and multimodal models. Learn with Jupyter notebooks, Google Colab and reusable Python modules.

An automated AI & ML news aggregator that clusters and ranks stories, generates AI briefings, and deploys a static news website twice daily using GitHub Actions.

Movie Recommendation System using Python and Machine Learning to recommend similar movies based on movie metadata and content similarity.

A 100% client-side, zero-latency local AI software architect powered by WebGPU and Next.js. Features live code rendering and VRAM auto-calibration

A simple neural network that recognizes handwritten 0s and 1s (MNIST, TensorFlow/Keras)

Hands-on implementations of ML, DL, NLP, and CV models and architectures from scratch and with modern frameworks.

Heard of minesweepers, right? It has numbers --> numbers, = computer language --> computer language = endless possibilities. Yeah so basically I am hiding message in minesweeper games

Advanced conversational RAG assistant with Hybrid Search, BM25, RRF, Multi-Query Retrieval, Contextual Compression, Memory, Structured Output, Evaluation, and Streamlit UI.

Benchmark framework for condition-sensitive reaction yield prediction, reaction condition cliffs, and OOD generalization.

Does a Dallas air quality model travel? Spatial transfer study across North Texas counties

Adversarial ML from scratch in NumPy: gradient-checked models, FGSM/PGD (targeted) attacks, adversarial training (35% -> 77% robust accuracy) and a manifold detector vs feature squeezing, with honest results.

California House Price Prediction using Machine Learning and Flask. Achieved an R² score of 82.91% and deployed on Render.

Telco churn prediction with Logistic Regression and Random Forest, plus a cost-based decision threshold

End-to-end multimodal ML system predicting YouTube breakout hits using LightGBM, PyTorch FusionNet, CLIP & MiniLM embeddings, SHAP explainability, and containerized FastAPI serving.

Data Science backtesting engine comparing XGBoost and Poisson models for Fantasy Premier League expected points (xP) prediction.

MCAP Studio is a workspace for turning MCAP sensor recordings into datasets for AI model training.

Frozen reproducibility benchmark for leakage-controlled flood-road inspection prioritisation across four Assam flood events.

Factor investing, portfolio optimization, and walk-forward backtesting using Indian equities in Python.

L2-regularized logistic regression implemented from scratch with NumPy, verified against scikit-learn

A machine learning project using Landsat satellite imagery to classify urban building ages. Explores feature engineering, model optimization, and cross-city transfer learning from Madrid to Amsterdam, with few-shot domain adaptation to improve generalization

Seven dry-bean varieties from real morphology features with duplicate-aware splits, calibrated SVM and held-out error analysis

AI study roadmap 2026: learn AI engineering by building — one track, stage by stage, zero spend on LLM APIs

Your AI data science team, with the receipts. Multi-agent analysis with deterministic verification and human review. Ask data questions in plain English; agents run the stats, a verifier checks the numbers, people decide.

Top Distributed Training Optimization (Opensource) Star if you like it!

A config-as-graph framework for deep learning: any pipeline, in any modality or domain, assembled from swappable components.