Arnav Gupta

I modernized Indian Railways’ FOIS Revenue Analytics Dashboard from a 6,000-line PHP codebase into modular FastAPI microservices, automating monthly freight reporting from 15+ hours to under 40 seconds.

I build end-to-end AI products, including an industrial predictive-maintenance platform that combines XGBoost, SHAP, LangChain, FAISS, SQLite, FastAPI, and Streamlit. Its failure-detection model achieved 98% accuracy, 0.98 ROC-AUC, 0.90 PR-AUC, and 87% recall on minority-class failures.

I also developed Silica-Sort, a Rust and Python hybrid external sorting engine that reached 87.6M elements per second and 253.6 MB/s for out-of-core sorting.

I've co-authored an IEEE INDIACom 2025 publication, won Horizon 1.0 at GDG On Campus VCET, and delivered Python applications on GCP for SNSS Global Services and YPP Technologies.