Harikrishnan Nair

Data Scientist @Ramp

CA, US

Designed and deployed scalable churn and fraud detection models using XGBoost, LSTMs, and Transformers, improving prediction accuracy by 22% and directly reducing customer attrition and revenue leakage.• Applied NLP techniques (BERT, Hugging Face Transformers, spaCy) to analyze millions of customer communication records, enhancing segmentation and personalization strategies, which improved campaign engagement rates by 25%.• Automated end-to-end ML pipelines with Python, Spark, Docker, and GitLab CI/CD, cutting deployment cycles by 40% while ensuring reliable and production-ready models across environments.• Integrated AWS Lambda, S3, and CloudWatch monitoring to deploy and track live models, reducing drift-related errors by 30% and improving system reliability for high-volume financial data processing.