Bharath Srinivas Ramanujan

Agentic AI & NLP Applied Scientist @ Five9

About

Data Scientist and ML Engineer with hands-on experience building end-to-end machine learning, deep learning, and generative AI systems. Built production-style ML pipelines across: • Credit Risk Modeling (AUC 0.9837 | KS 85.99 | 94% recall for defaulters) • Segmented Healthcare Premium Prediction (XGBoost R² up to 0.9971) • Computer Vision damage detection using transfer learning (78% validation accuracy) • Retrieval-Augmented Generation (RAG) systems reducing LLM API cost by ~70% Strong foundation in: Python, SQL, Scikit-learn, XGBoost, PyTorch, Power BI, Optuna, Model Evaluation (ROC-AUC, Gini, KS), Feature Engineering, Cross-Validation. Experience validating production ML pipelines and improving CI/CD workflows (GitHub Actions, MLflow, AWS SageMaker exposure). Focused on building statistically sound, business-aligned ML systems rather than purely academic models. Actively seeking entry-level Data Scientist, ML Engineer, and Data Analyst roles.

Country

City

United States

Industry

Computer Software

Skill

AWS SageMaker, MLflow, ChromaDB, LangChain, Retrieval-Augmented Generation (RAG), Hyperparameter Tuning (Optuna), Model Evaluation (ROC-AUC, KS, Gini), Feature Engineering, DAX, Microsoft Power BI, Deep Learning, PyTorch, XGBoost, Logistic Regression, SQL, Python (Programming Language), MLOps, Machine Learning, Scikit-Learn, Data Visualization

Experience

Agentic AI & NLP Applied Scientist

Five9

2026-6 - Present · 4 mos

California, United States

Machine Learning Engineer

Vosyn

2025-9 - 2025-11 · 3 mos

Etobicoke, ON

Validated end-to-end ML/data pipelines to ensure correctness of feature engineering, transformations, and model outputs prior to release. • Designed structured validation test cases aligned with acceptance criteria across staging and production environments. • Improved CI/CD workflows using GitHub Actions by reducing hard-coded dependencies and increasing deployment configurability. • Collaborated in Agile sprints to operationalize ML validation processes for production-grade systems.

Data Scientist

AtliQ Technologies

2025-6 - 2025-8 · 3 mos

·Developed a logistic regression–based credit risk model for default prediction on an imbalanced financial dataset, achieving AUC 0.9837 and KS statistic 85.99. ·Performed hyperparameter optimization using Optuna and RandomizedSearchCV to improve model stability and predictive performance. ·Built healthcare premium prediction models using XGBoost, achieving R² up to 0.9971 through feature engineering and cross-validation. ·Conducted multicollinearity diagnostics (VIF), feature selection, and model evaluation to ensure reliable predictors and reduce overfitting. ·Communicated model performance using ROC curves, decile analysis, and business impact interpretation to translate technical findings into actionable insights. ·Gained exposure to ML experimentation workflows including MLflow tracking and AWS SageMaker deployment environments