I transitioned into Data Science from a mechanical engineering background, and it shows in how I work — I care as much about whether a system holds up under real conditions as I do about model accuracy. I build things end-to-end: data pipelines, model training, serving infrastructure, and the deployment plumbing that makes it all actually run in production, not just in a notebook.
- 🔭 Currently building production-grade MLOps pipelines with reproducible, containerized deployments
- 🌱 Learning by shipping — every project goes through CI/CD, containerization, and real deployment, not just
model.fit() - 🐛 I enjoy debugging production issues almost as much as building the feature in the first place
- 📫 Reach me at krvipin15@tutamail.com
Production-grade MLOps pipeline for predicting customer churn — from data ingestion to a live, containerized serving stack.
- 🧠 XGBoost model achieving 0.92 ROC-AUC on 594K+ customer records, with threshold tuning that raised churner recall from 64% → 78.5%
- ⚙️ DVC-orchestrated pipeline (ingest → validate → preprocess → validate → train) with Pandera schema validation
- 🚀 FastAPI inference service + Streamlit dashboard, containerized with Podman and published to GHCR
- ✅ CI/CD with CodeQL, OSV-Scanner, Dependabot, and 23 pre-commit hooks across 12 tools
- 📖 Full documentation site with API reference auto-generated from docstrings
Open to Software Engineering / Data Science internship and entry-level opportunities.