Data Scientist Senior

Name: Vinay

Senior Data Scientist/ GenAI Architect

Phone:323-***-****

Email: *************@*****.***

Professional Summary:

Highly accomplished Senior Data Scientist with 10+ years of progressive experience specializing in end-to-end Generative AI/LLM solution development, MLOps architecture, and advanced Risk Modeling across the Healthcare and Financial Services sectors.

Proven expertise in leveraging Google Cloud Platform (GCP/Vertex AI) and AWS ecosystems to design, deploy, and govern high-impact, scalable machine learning systems.

Expert in designing and deploying production-grade GenAI workflows on GCP using Gemini, LangChain, and LangGraph for use cases like clinical note summarization, compliance validation, and conversational AI.

Architected and deployed high-performance Retrieval-Augmented Generation (RAG) pipelines using Vertex AI Vector Search to enable real-time similarity searches over millions of clinical documents.

Managed end-to-end ML and GenAI lifecycle using Vertex AI Pipelines (CI/CD) and AWS SageMaker Pipelines, ensuring automated retraining, validation, and compliant deployment.

Deep hands-on experience across major cloud environments, including GCP (Vertex AI, BigQuery) and AWS (SageMaker, S3, EKS), for scalable data processing and model serving.

Expertise in developing and deploying complex time-series forecasting models (e.g., credit card default rates) and Gradient Boosting Machines (XGBoost, LightGBM) for business-critical risk assessment.

Proficient in analyzing petabyte-scale datasets, performing feature engineering, and optimizing large-scale data pipelines using tools like DuckDB, PySpark, Snowflake, and BigQuery.

Implemented robust model monitoring (Splunk) and governance practices, detailing statistical metrics (AUC, Gini) and explainability (SHAP values) for regulatory review.

Production-level proficiency in Python, TensorFlow, Scikit-learn, Docker, Kubernetes, and building real-time applications with Streamlit and Flask/FastAPI.

Successfully embedded AI/ML models into patient-facing, physician, and business applications (e.g., ServiceNow integration), providing real-time decision support and driving substantial business growth (achieved millions in revenue through deployed AI solutions).

Technical Skills:

Cloud & DevOps

Microsoft Fabric, OneLake, Azure Synapse, Data Lake, Microsoft Excel, AWS (SageMaker, S3, EKS, Redshift, Glue), GCP (Vertex AI, BigQuery), Azure, Docker, Kubernetes, OpenShift, Git, Terraform.

Data & GenAI Tools

LangChain, LangGraph, LangSmith, FAISS, Pinecone, Ray, MLflow, dbt, Apache Spark, Kafka

Machine Learning

Linear Regression, Multivariate Linear Regression, Logistic Regression, Gradient Descent Linear, Regression, Discriminant Analysis, Naive Bayes, K-Nearest Neighbor (KNN), Learning Vector Quantization (LVQ), Support Vector Machine (SVM), Bagging, Random Forest, Boosting, AdaBoost, Gradient Boosting Machine (GBM), XGBoost, LightGBM, CatBoost, Neural Network, Natural Language Processing (NLP)

Python Libraries

Python (Primary), SQL, PySpark, R, Unix/Linux Shell.

Deep Learning

Artificial Neural Networks, Convolutional Neural Networks, RNN, Deep Learning on AWS, Keras API.

Data Visualization

Tableau, Python (Matplotlib, Seaborn), R(ggplot2), Power BI, QlikView, D3.js, Rshiny

Programming

Python, PySpark, Java, C#, T-SQL, SAP ABAP, AZURE PowerShell, R, TensorFlow, PyTorch, LLMs

Data Sources

PostgreSQL, MongoDB, MySQL, HBase, Amazon Redshift, snowflake, Neo4J, SQL Server, Oracle, PostgreSQL, Snowflake, Cosmos DB, HBase, Cassandra.

Other tools

TensorFlow, Keras, AWS ML, NLTK, SpaCy, Gensim, MS Office Suite, GitHub, AWS(EC2/S3/Redshift/EMR/Lambda/Snowflake), GCP, Azure, Bigquery, Stack driver, Fire base, Prometheus, Hadoop, Hive, HDFS, Apache Kafka, Apache Spark, Data Bricks, Teradata, Sqoop, ClinicalBERT, BioGPT, Med-PaLM

Professional Experience:

Centene Corporation - St Louis, MO March 2024 to Present

Senior Data Scientist

Responsibilities:

Designed and deployed robust GenAI automation workflows on GCP using LangChain and LangGraph with Gemini for dynamic clinical note summarization, compliance document validation, and adverse event reporting in EHR systems.

Engineered and optimized prompts for the Gemini model family within the Vertex AI platform, utilizing zero-shot, few-shot, and CoT prompting for use cases like automated pre-authorization, medical coding validation, and clinical documentation improvement (CDI).

Architected and deployed high-performance Retrieval-Augmented Generation (RAG) pipelines using Gemini and Vertex AI Vector Search, enabling real-time similarity searches over millions of clinical guidelines and patient records.

Developed high-performance AI-powered applications with Streamlit and Flask/FastAPI, integrating Gemini for real-time evidence-based clinical Q&A and diagnostic insight generation.

Leveraged Vertex AI and cloud-based HPC clusters for parallel processing and distributed fine-tuning of LLMs on large-scale medical texts, significantly reducing training time for real-time decision support.

Optimized generative models within the Vertex AI ecosystem for image-to-text synthesis (e.g., radiology reports from scans) and summarization, enhancing pathology and claims document workflows.

Integrated Gemini models with ServiceNow workflows to automate patient intake document routing, flag anomalies like medication conflicts, and ensure compliance across hospital operations.

Developed advanced conversational AI chatbots and virtual assistants using frameworks suitable for Vertex AI Agent Builder, enabling secure patient triage and scheduling.

Spearheaded initiatives for chronic disease management by transforming manual patient monitoring into AI-powered solutions using Gemini-based workflows for care coordination.

Developed GenAI-driven personalized patient education algorithms on GCP, tailoring discharge instructions and follow-up care recommendations to individual health literacy levels.

Managed end-to-end ML workflows on GCP, using Vertex AI Pipelines for CI/CD to automate model training, validation, and secure deployment, ensuring continuous improvement and compliance.

Built, tuned, and deployed machine learning models (regression, forecasting) using Vertex AI Training to forecast patient readmission risk and predict disease progression, supporting dynamic resource allocation.

Developed and optimized ML algorithms using Python, TensorFlow, and Scikit-learn; deployed scalable, HIPAA-compliant models on GCP leveraging Vertex AI Prediction.

Implemented robust monitoring using tools like Splunk to track critical Vertex AI model metrics (accuracy, latency, data drift, bias) to maintain fairness and reliability in clinical applications.

Automated data ingestion and preprocessing pipelines by integrating Vertex AI with secure GCP data sources like BigQuery and Google Cloud Storage to prepare data for model training and analysis.

Continuously debugged, retrained, and improved model performance within the Vertex AI MLOps framework, using fresh clinical data and feedback loops to maintain high accuracy in diagnostic support systems.

Analyzed large volumes of structured and unstructured patient data (EHRs, pathology reports, genomic data) from sources analogous to BigQuery, performing feature engineering to extract actionable insights for personalized medicine.

Architected and maintained large-scale data pipelines on GCP, storing secure AI/ML datasets in databases like BigQuery for near-real-time clinical and research analytics.

Built and presented actionable clinical and operational insights through Tableau dashboards, supporting data-driven decisions on hospital efficiency and patient care quality.

Collaborated with cross-functional teams (Physicians, IT Security) to embed Gemini-powered models into patient-facing and physician applications, delivering real-time decision support.

Environment: GCP (Google Cloud Platform), Vertex AI, Gemini, LangChain, LangGraph, Vertex AI Vector Search, BigQuery, Google Cloud Storage, Python, TensorFlow, Scikit-learn, Streamlit, Flask, FastAPI, ServiceNow, Splunk, Tableau.

CapitalOne - Mclean, VA Sept 2022 to Feb 2024

Senior Data Scientist

Responsibilities:

Designed a multi-horizon time-series forecasting model architecture to predict credit card default rates, incorporating complex macroeconomic indicators and internal consumer behavior data features.

Analyzed petabyte-scale structured datasets stored in AWS S3 using advanced sampling techniques to build unbiased training sets for low-incidence risk modeling problems (e.g., severe delinquency).

Formulated multi-step analytical hypotheses and reasoning-based questions to test the statistical significance of new feature engineering pipelines on model predictive power, specifically on cloud-ingested data.

Executed deep data exploration on customer transaction histories to identify and validate novel features, demonstrating their high Information Value (IV) for credit risk scoring models.

Architected a segmentation strategy based on unsupervised machine learning (K-Means Clustering) to group customers for tailored loan offers, resulting in a reduction in acquisition cost.

Wrote production-ready Python scripts, containerized using Docker, to train and serialize a suite of Gradient Boosting Machine (GBM) models, facilitating seamless deployment to the cloud.

Utilized DuckDB's vectorized processing engine and its Python API to efficiently execute large-scale, complex SQL feature calculation queries directly on Parquet files stored in AWS S3.

Developed and implemented an automated CI/CD pipeline for ML model retraining and deployment using AWS SageMaker Pipelines, cutting model update latency

Managed the end-to-end versioning of Python modeling notebooks and associated SQL feature extraction code using GitHub, integrating repository hooks for MLOps validation checks.

Constructed parameter-tuned SQL queries using temporary tables and advanced aggregations to extract monthly snapshot features, serving as the reliable source for cloud-based model training jobs.

Implemented a Model Registry on a cloud platform (e.g., AWS SageMaker) to catalog and track metadata, performance metrics, and governance approval for over 15 distinct credit risk models.

Automated the scaling of cloud compute resources (AWS EC2) via Python scripts to efficiently handle the varying demands of daily model inference jobs, minimizing operational cost.

Produced clear, human-readable model performance reports, detailing the statistical reasoning behind the chosen metrics (e.g., AUC, Gini coefficient) and their implications for business strategy.

Delivered model deployment specifications by formatting the output of model results and key explainability metrics (e.g., SHAP values) into a uniform, structured JSON format for API consumption.

Authored comprehensive documentation within Jupyter Notebooks, explaining the mathematical and computational complexity of advanced model fitting procedures to internal review boards.

Interpreted model residual analysis and diagnostics to identify systemic biases, translating complex statistical findings into clear recommendations for data scientists and line-of-business stakeholders.

Refined the JSON output schema to include model metadata, training hyperparameters, and full data lineage, optimizing the monitoring and governance of deployed ML models.

Validated the stability and bias of newly trained models by running thorough cross-validation routines and challenger/champion tests on a dedicated AWS EMR cluster for large-scale processing.

Provided in-depth reviewer feedback on the statistical rigor of peer model validation plans, focusing on correct application of out-of-time (OOT) testing and feature drift monitoring in the production environment.

Leveraged iPython interactive sessions for rapid exploratory model fitting and feature set evaluation before committing logic to the automated MLOps workflow.

Collaborated with platform engineers to configure cloud infrastructure (AWS Lambda and API Gateway) for low-latency, real-time model serving.

Ensured model quality post-deployment by implementing automated data quality checks and model performance monitoring within the cloud environment.

Environment: Python, PySpark, FastAPI, PyTorch, TensorFlow, Scikit-learn, XGBoost, LightGBM, CatBoost, AWS (EKS, SageMaker), GCP (Vertex AI, AI Platform, BigQuery), Kubernetes, Docker, Terraform, Jenkins, MLflow, Airflow, dbt, Evidently AI, SHAP, LIME, Tableau, Power BI, Snowflake, SQL, Agile, CI/CD.

Vanguard - Valley Forge, PA June 2020 to Aug 2022

Data Scientist

Responsibilities:

Translate complex business challenges into mathematical and statistical inquiries, leading the design, development, and implementation of ML/AI solutions that drive substantial business impacts.

Spearheaded the development of machine learning (ML) and AI models to enable data-driven decision-making and support Vanguard's wealth management advice service.

Led the design, development, and implementation of ML solutions resulting in a 130% increase in Personal Advice Service (PAS) growth.

Achieved a remarkable milestone by generating over millions in revenue through the deployment of AI solutions.

Led the design, development, and deployment of hyper-personalization systems for millions of investors using reinforcement learning algorithms.

Developed sophisticated recommended systems for clients and advisors, significantly improving conversion rates and reducing attrition.

Designed ML models tailored for Retail Investor Group (RIG) marketing initiatives to create targeted and effective campaigns.

Enhanced customer experience by providing precise forecasts of phone call demand and recommending optimal staffing strategies.

Leveraged cutting-edge Large Language Model (LLM) technologies to proactively identify instances of fraud and cognitive decline, enhancing risk mitigation.

Mentored and supervised a team of data scientists throughout the entire data science project life cycle.

Empowered the entire team through comprehensive Large Language Model (LLM) training.

Conducted extensive research into emerging technologies and AI algorithms (Optimization, Deep Learning, Reinforcement Learning) with the potential to revolutionize the financial industry.

Played a pivotal role in transitioning analytics teams from traditional platforms (SAS, desktop R/Python) to a cutting-edge big data and unified analytics platform on the Cloud.

Demonstrated the predictive power of machine learning algorithms and big data platforms to internal peers and clients to showcase potential and drive adoption.

Environment: Machine Learning, Python, Pandas, Numpy, Matplotlib, Scikit-learn, Tableau, Hadoop, YARN, Spark, GCP, 3NF, flume, UNIX, Zookeeper HBase, Kafka, NoSQL, Cassandra, Elastic Search, Sqoop.

American International Group - New York, NY Jan 2018 to May 2020

Data Scientist

Responsibilities:

Actively involved in designing and developing data ingestion, aggregation, and integration in the Hadoop environment.

Developed Sqoop scripts to import export data from relational sources and handled incremental loading on the customer, transaction data by date.

Performed data analysis and data profiling using complex SQL queries on various sources systems including Oracle 10g/11g and SQL Server 2012.

SQL/Transact SQL server DBA support: Strong development skills in coding SQL stored procedures for premium and claims audit issues.

Finding bogus claims, finding audit trail issues is SQL database, finding data loading issues on SQL server.

MS Access developer. Built a normalized relational database in MS Access for new project. 10,000 record databases needing import/export data capabilities as well as reporting.

Designed, trained, and deployed machine learning models using SageMaker's built-in algorithms and custom solutions.

Provide guidance and supervision to teams as they carry out work activities. Recruits and develops staff within area of responsibility in accordance with policy and manage vendors engaged in big data analytics.

Influences individuals within and outside the department. Prepares and presents reports to all levels of leadership and staff.

Built end-to-end ML pipelines with SageMaker Pipelines to automate data preprocessing, model training, and deployment.

Developed predictive models using Python and scikit-learn to support business decision-making in customer segmentation.

Applied statistical testing techniques such as hypothesis testing and confidence intervals to validate product changes.

Built and trained machine learning models for classification and regression tasks using libraries like TensorFlow, XGBoost, and LightGBM.

Actively involved in A/B tests, defined metrics to validate user interface features, calculating sample size and checking statistical assumptions for tests.

Carrying out specified data processing and statistical techniques such as sampling techniques, estimation, hypothesis testing, time series, correlation and regression analysis Using R.

Environment: R, SQL server, Oracle, HDFS, HBase, MapReduce, Hive, Impala, Pig, Sqoop, NoSQL, Tableau, RNN, LSTM, Unix/Linux, Core Java

Hitachi Solutions India Pvt. Ltd., India July 2015 to Sept 2017

Data Analyst

Responsibilities:

Established management reporting capabilities, crafting dashboards for senior management to drive data-driven decision-making.

Developed end-to-end machine learning solutions for time-series KPI values and predictive analytics.

Utilized Prophet, LSTM, TBATS, VAR, VARMAX, Random Forest Regression, and other models for time series analysis and market ranking prediction.

Cleaned, transformed, and validated structured data from multiple sources to generate actionable business insights.

Conducted exploratory data analysis (EDA) using Python (Pandas, NumPy, Seaborn) to uncover trends and anomalies.

Wrote optimized SQL queries and stored procedures to retrieve and aggregate data from PostgreSQL and Snowflake databases.

Designed custom performance metrics and KPIs to track product usage, customer engagement, and feature adoption trends.

Analyzed multi-channel marketing data to attribute customer acquisition costs (CAC) and optimize ROI across digital platforms.

Conducted cohort and funnel analysis to evaluate user retention over time and identify drop-off points in the user journey.

Performed A/B testing and multivariate experiments using t-tests, ANOVA, and chi-square to evaluate feature impact and validate hypotheses.

Created statistical models using linear regression, logistic regression, and clustering techniques to support business decisions.

Estimated confidence intervals and p-values to assess the significance of product or marketing changes.

Visualized KPIs and trends with clear charts, graphs, and filters, enabling non-technical users to explore insights independently.

Used Looker and Tableau to track KPIs for sales, marketing, and product, enabling real-time business decision-making.

Extracted and transformed data using dbt (data build tool) to support scalable reporting infrastructure.

Implemented data quality checks and documentation to ensure accuracy, consistency, and stakeholder trust.

Presented findings to non-technical stakeholders using storytelling and visualization to support data-driven decisions.

Employed Dynamic Time Warping, K-means Clustering, Hierarchical Clustering, and other algorithms for event detection and pattern recognition in alarm data.

Utilized PySpark for preprocessing and analyzing large-scale alarm datasets, facilitating efficient data cleaning and feature engineering.

Environment: React, Angular, jQuery, HTML, CSS MongoDB, Redis, Tableau, AWS Services (S3, EC2, Lambda, CloudTrail, CloudWatch, SQS, SNS)

Contact this candidate