Prabhav Khare

I’m a Machine Learning Research Intern at NIT Rourkela, currently pursuing a Bachelor of Technology in Chemical Engineering (CGPA: 6.6/10) from the National Institute of Technology, Rourkela. My focus is turning data into reliable predictions and usable tools for real users.

In my internship, I automated Python ETL pipelines across RCSB, UniProt, and ChEMBL APIs for 10,000+ compounds, reducing preprocessing time by 2 hours per experiment cycle. I built an ensemble (Random Forest + DNN) for pIC50 bioactivity prediction on 8,825 samples, reaching R = 0.79 2, and tuned models with GridSearchCV and MLflow tracking across 48 hyperparameter combinations.

I also shipped end-to-end workflows: I deployed a virtual screening pipeline (AutoDock Vina + 10ns MD simulations) via Streamlit on Render, enabling one-click candidate evaluation for non-technical researchers. Across projects, I enjoy designing data systems that support confident analytics—from simulation-based testing to fast reporting.

Beyond work, I’ve built retail and geospatial intelligence systems using LangChain RAG, pgvector, FastAPI/Streamlit, and TimescaleDB SQL, and I contribute to open learning as a Kaggle Dataset Contributor (10+ open-source datasets). I’m energized by practical experimentation—like A/B testing, statistical validation, and ensemble modeling—to deliver measurable impact.