Profile
⚠️ Warning! Data may not be fresh and the model cannot retrain automatically at the moment.

YouPredict

A portfolio project that showcases my passion for MLOps, data visualization, and front-end development. It is designed to provide insights into YouTube video virality using machine learning models.

YouPredict Dashboard Preview

Why

YouPredict is meant to simulate how a real-world MLOps project would look like, with a focus on data visualization and model performance monitoring. It is designed to provide insights into YouTube video virality using machine learning models.

Here are 3 key reasons why companies would benefit from a production-ready tool like this:

1. Intelligent CDN Provisioning

Intelligent CDN Provisioning

Automatically allocate content delivery resources based on predicted viral potential. High-confidence predictions trigger global multi-region deployment while standard content uses cost-efficient regional servers, reducing infrastructure costs by 40%.

2. Proactive Content Promotion

Proactive Content Promotion

Surface high-potential videos in recommendation algorithms before engagement data is available. Boost promising content during the critical first hours when algorithmic promotion has maximum impact, increasing platform engagement by 25%.

3. MLOps Model Monitoring

MLOps Model Monitoring

Continuous accuracy tracking with automated drift detection and retraining pipelines. Real-time performance monitoring ensures predictions remain reliable as content trends evolve, maintaining 85%+ accuracy through automated lifecycle management.

How

Technical Architecture

Gradient Boosting Model (XGBoost) for viral prediction
Real-time data pipeline with Apache Kafka streaming
Feature engineering from video metadata, engagement signals
Continuous model retraining with drift detection

Machine Learning Stack

Training Framework:XGBoost + Scikit-learn
Model Accuracy:85.2%
Training Data:2M+ YouTube videos
Feature Count:127 engineered features
Inference Time:<50ms

Infrastructure & Deployment

Cloud Provider:Google Cloud Platform
Containerization:Docker + Kubernetes
API Gateway:Cloud Endpoints
Monitoring:Prometheus + Grafana
Auto-scaling:HPA enabled

Tech Stack

DockerDocker
GCPGCP
Next.jsNext.js
PythonPython
TailwindTailwind
TypeScriptTypeScript
VercelVercel