Showing 30 of 1,478 · Category: Machine Learning
Top Multi Cloud ML Orchestration (Opensource) Star if you like it!
Predictive ML-Driven LLM Routing, Token FinOps & Data Drift Observability Platform built with XGBoost, Scikit-Learn, FastAPI, FastMCP, and Streamlit.
Execution-based code repair benchmark with controlled error generation, QLoRA training, and sandboxed evaluation.
Buy now or wait? Predicting flight price drops
Ryan Vo | Work in progress: sentinel-lab-ryan-vo
Top Human In The Loop Machine Learning (Opensource) Star if you like it!
Predicting customer complaint escalation with ML, Azure AI Language sentiment analysis and SHAP explainability, plus a fairness audit and Power BI dashboard, for a fictional UK financial services firm.
End-to-end fraud detection ML system with leakage-aware feature engineering, FastAPI inference, Docker, testing, and CI.
Confidence-gated GitHub issue triage powered by a fine-tuned Laya decision model
Real-time space weather intelligence platform with ML-powered solar storm forecasting.
Tiny open-source model (1.2 MB) that understands percentage questions in English, Portuguese and Spanish and returns the right calculation. Runs in the browser in ~1 ms.
Prácticas de Aprendizaje Automático I (Grado en Ciencia e Ingeniería de Datos, UAM): preprocesado, regresión lineal/Lasso/Ridge, clasificación con regresión logística y Naïve Bayes implementado desde cero, y PCA. Python, pandas y scikit-learn.
PortfolioLens AI combines Python, SQL, and machine learning to analyze synthetic loan portfolios. Features data-quality controls, delinquency and roll-rate reporting, an evidence-based AI copilot, an interactive dashboard, model evaluation, automated tests, and a 28-second demo.
Python ML pipeline using Isolation Forest to detect unusual system logs, prioritize anomalies by severity, and provide an interactive Streamlit dashboard with CSV/JSON exports. Includes Docker containerization, health checks, automated tests, and GitHub Actions CI with container smoke testing.
Learning to rank for Expedia hotel search with LightGBM LambdaMART ensembles
Predicts the finishing order of an F1 Grand Prix from pre-race data (FastF1, scikit-learn, XGBoost), with a Streamlit app.
End-to-end Computer Vision pipeline for employee activity recognition and productivity analysis using YOLO11, BoT-SORT, SlowFast, MediaPipe, and multi-model fusion.
Top Data Labeling Quality Control Systems (Opensource) Star if you like it!
Phishing email detection with machine learning and explainable risk scoring (Python, scikit-learn, Streamlit)
ML inflow forecasting + multi-objective optimization (NSGA-II) of Vietnam’s Hoa Binh reservoir: +7% hydropower and −20% flood risk in Hanoi vs. the current policy.
Virtual Lab JST berbasis web (SPA) untuk simulasi pelatihan dan evaluasi JST pada data tabular nyata (CSV custom/Kaggle), bukan sekadar gerbang logika biner.
2 LLMs designed and trained entirely by Opus 5.5
Sentiment tools/vendors that fail on sarcasm, slang, and emoji.
Explainable AI for detecting anomalous access to electronic health records using contextual behavioural analytics and privacy-safe synthetic data.
Which fire is it? Labels every NASA FIRMS hot spot over India as gas flare, industry or farm burning, with a research console, the AgniNetra Live public app, and the open BharatThermal-1 dataset of 601,941 detections.
Perovskite open-data ML: 42k-device process-PCE regression with confounder ablation, Bayesian optimization, YOLOv8 defect detection
Investigator dashboard for detecting and reporting fake social media profiles — ML classifier + LLM content analysis + investigator UI. Frontend complete; backend/model integration in progress.
Fraud Triage & Siskamling Digital — public smishing report & check service for Indonesia. Calibrated ensemble + deterministic rule layer, evaluated with pre-registered gates; results published as-is, including failures.
LiquidDynamo transforms your MacBook’s notch into an interactive command center with fluid spring physics and liquid-glass visuals. Control media, interact with 3D devices, bridge Android, accelerate downloads, monitor system telemetry, and stage files—or enjoy a dynamic island on external displays and non-notched Macs.
Image classifier for 6 scene classes (buildings, forest, glacier, mountain, sea, street) built with PyTorch transfer learning. Fine-tunes a pretrained ResNet-18 on the Intel Image Classification dataset with data augmentation, reaching 89.6% test accuracy. Includes a predict script and Gradio web demo.