Balaji Chinthagunta

AI/ML Engineer with over five years of experience building machine learning and NLP systems across banking, insurance, and engineering. Currently developing a production RAG search platform and multi-step agent workflows at U.S. Bank that let risk and compliance teams query policy and regulatory documents directly.

Earlier work covered claims analytics at ICICI Prudential and predictive maintenance models at Cyient. M.S. in Data Science from University at Albany. AWS Certified AI Practitioner and Microsoft Azure AI Engineer Associate.

GenAI / AI-ML Engineer - U.S. Bank - New York, NY

(2025-05)

• Built and shipped a RAG-based search tool that lets risk and compliance analysts query policy and regulatory documents instead of reading them by hand; the system is now used daily by the compliance team.
• Designed the full document ingestion pipeline (parsing, chunking, embeddings, and metadata storage) so results can be filtered by policy type and effective date.
• Implemented LangGraph multi-step agent workflows that call internal tools, keep track of conversation state, and route high-risk answers to a human reviewer before release.
• Developed FastAPI and Spring Boot services that connect the tool to other bank applications, with input validation and structured logging that make production issues easier to track down.
• Set up prompt versioning and a golden test set of questions with expected answers; the team runs these checks before any prompt or model change goes live.
• Fine-tuned smaller open-source models with SFT and DPO to classify the intent of compliance queries, then served them with vLLM.
• Built the React front end for search and policy Q&A and shipped updates every sprint with product, architecture, and QA.

AI/ML Engineer - ICICI Prudential - India

(2022-01 - 2023-07)

• Automated the daily claims data preparation process in Python and SQL, replacing a manual spreadsheet workflow that used to take several hours each morning.
• Applied NLP to claim descriptions and adjuster notes to classify issue types and surface recurring problems for the claims review team.
• Wrote SQL validation checks with joins, CTEs, and aggregations that caught duplicate claims, missing policy fields, and payment mismatches before reports went out.
• Prepared the datasets and KPI logic that powered Tableau and Power BI dashboards used in weekly and monthly claims reviews.
• Helped build policyholder segments from demographics, premium history, and claim frequency for renewal and retention analysis.

Data Scientist - Cyient - India

(2019-01 - 2021-12)

• Built classification, regression, and forecasting models with scikit-learn, TensorFlow, and PyTorch to predict maintenance needs and flag equipment at risk of failure.
• Cleaned and validated engineering, asset, and inspection records in Python and SQL so the data was ready for modeling.
• Grouped inspection notes by text similarity to find repeat issues, giving field teams a clearer list of what to fix first.
• Compared models on precision, recall, F1, RMSE, and MAE and documented where each approach performed well and where it fell short.
• Created Tableau and Power BI dashboards on asset trends and automated the weekly and monthly reports that used to be built by hand.

Master of Science - Data Science - University at Albany (2023-08 - 2025-05)

Bachelor of Science - Sri Harshini Degree College (2016-03 - 2019-03)