Harinath Chakali

Ai Ml Engineer @Citizens Bank

US

Worked on Citizens Bank’s AI-driven Customer Experience Platform aimed at enhancing personalized banking interactions and financial advisory services via natural language understanding and recommendation systems, collaborating closely with product managers, data scientists, and requirement gathering teams to define project scope and deliverables. Built scalable ingestion pipelines using Python (Beautiful Soup, Pydantic), SQL Alchemy, and AWS Glue to process millions customer interaction logs, transaction histories, and product usage data stored in PostgreSQL with clear data lineage for regulatory audits. Engineered user behavior and interaction features such as intent classification, sentiment scores, and transaction patterns using pandas, spaCy, and custom rule-based extractors, improving model input quality by 27%. Designed and fine-tuned transformer-based NLP models (BERT, RoBERTa) on AWS Sage Maker for intent recognition, query classification, and personalized product recommendation, boosting model accuracy from 0.78 to 0.90 with domain-specific training. Utilized LoRA and PEFT techniques with Hugging Face Accelerate and AWS EC2 instances to efficiently fine-tune models on proprietary customer datasets while adhering to Citizens Bank’s privacy and data governance policies (SOC 2, PCI DSS). Conducted extensive hyperparameter optimization using Optuna with Bayesian search across 100+ experiments on AWS Sage Maker, applying early stopping and k-fold cross-validation, while tracking results through MLflow and GitLab CI/CD.

University of North Texas

Master's degree

Geethanjali College of Engineering and Technology

Bachelor of Technology - BTech