harsh = {
"role": "Machine Learning Engineer",
"focus": ["Production ML Systems", "MLOps", "Cloud Deployment"],
"stack": ["Python", "Docker", "AWS", "FastAPI", "TensorFlow", "PyTorch"],
"education": "B.Tech Computer Engineering @ NIT Delhi (GPA: 8.5/10)",
"currently": "Building end-to-end ML pipelines for real-world impact",
"fun_fact": "I turn raw data into business value 🚀"
}ML & Data Science
MLOps & Cloud
Data Engineering
Real-time air quality forecasting with live API data ingestion, automated CI/CD, and Streamlit deployment
- Automated hourly data collection pipeline via GitHub Actions
- Full Docker + docker-compose containerization for reproducibility
- Live deployed on Streamlit Cloud
End-to-end ML platform with ensemble models achieving 89% ROC-AUC, deployed on AWS
- Ensemble: Logistic Regression + Random Forest + XGBoost → 89% ROC-AUC
- < 100ms inference latency at 1,000 requests/min on AWS SageMaker
- 30+ engineered features boosting performance by 12%
Data pipeline + regression models that identified $1.24M in cost optimization opportunities
- Processed 10,000+ records with scalable Python/SQL pipeline
- Identified $1.24M in optimization opportunities
- Dashboards showing $186K savings from 15% inventory reduction
- 🎖️ IBM AI Developer Professional Certificate
"Turning data into decisions, models into products."