Generative AI & LLM Engineer

VASAVI KOTTAKOTA

Generative AI Engineer AI/ML Engineer Data Scientist LLM Applications RAG MLOps

+1-469-***-**** *****************@*****.*** Linkedin

Professional Summary:

·Generative AI Engineer / AI-ML Engineer / Data Scientist with 8+ years of experience designing, developing, deploying, and supporting enterprise-grade Generative AI, LLM applications, RAG pipelines, Machine Learning, Deep Learning, NLP, Computer Vision, and data-driven solutions across banking, healthcare, insurance, and technology consulting domains.

·Strong hands-on experience building production-ready LLM-powered applications, including enterprise assistants, intelligent document processing systems, mortgage underwriting assistants, clinical decision support tools, semantic search platforms, conversational AI systems, and AI-powered workflow automation.

·Experienced in designing end-to-end Retrieval-Augmented Generation architectures, including document ingestion, OCR/text extraction, parsing, chunking, embedding generation, vector indexing, semantic retrieval, hybrid search, reranking, prompt orchestration, grounded response generation, citation handling, hallucination reduction, and LLM evaluation.

·Hands-on experience with Agentic AI and multi-agent workflows using LangChain, LangGraph, LlamaIndex, tool calling, function calling, agent orchestration, planning workflows, retrieval agents, validation agents, and human-in-the-loop review workflows.

·Strong experience working with commercial and open-source LLMs including GPT-4, Claude, LLaMA, BERT, RoBERTa, T5, BioBERT, ClinicalBERT, and Hugging Face Transformers, with practical exposure to prompt engineering, prompt optimization, fine-tuning, LoRA, QLoRA, PEFT, SFT, LLM evaluation, and domain-specific model adaptation.

·Proven background in Machine Learning and Deep Learning, including classification, regression, clustering, anomaly detection, fraud detection, risk scoring, churn prediction, recommendation systems, forecasting, predictive modeling, NLP, computer vision, OCR, and document intelligence using Python, Scikit-learn, TensorFlow, PyTorch, XGBoost, LightGBM, Pandas, NumPy, Spark, and PySpark.

·Experienced in developing NLP and document intelligence systems for mortgage documents, underwriting policies, clinical notes, claims documents, insurance forms, customer feedback, policy documents, and enterprise knowledge bases using BERT, LayoutLM, Donut, spaCy, NLTK, OCR pipelines, embeddings, and LLM-based extraction workflows.

·Strong cloud AI experience across AWS, Azure, and GCP, with AWS-focused banking and insurance AI platforms, Azure-focused healthcare GenAI and clinical NLP solutions, and GCP-focused early analytics, NLP, computer vision, and data processing workflows.

·Skilled in building scalable backend and AI services using FastAPI, Flask, REST APIs, microservices, Docker, Kubernetes, AWS EKS, Azure AKS, CI/CD pipelines, GitHub Actions, Jenkins, MLflow, DVC, Airflow, and model-serving workflows.

·Experienced in implementing MLOps and LLMOps practices, including model versioning, prompt versioning, experiment tracking, model registry, automated deployment, CI/CD for ML, drift detection, model monitoring, LLM evaluation, latency tracking, hallucination testing, token usage monitoring, and production release governance.

·Strong background in processing large-scale structured and unstructured datasets using Python, SQL, PySpark, Apache Spark, Hadoop, Airflow, data lakes, cloud storage, ETL pipelines, data validation workflows, feature engineering pipelines, and distributed data processing frameworks.

·Experienced in building explainable and responsible AI solutions using SHAP, LIME, fairness assessment, disparate impact analysis, bias mitigation, PII redaction, content safety guardrails, audit logging, role-based access control, and secure AI deployment practices.

·Skilled in developing business-facing dashboards and analytics solutions using Power BI, Tableau, Streamlit, Plotly, Matplotlib, and Seaborn to communicate model performance, risk trends, fraud indicators, clinical insights, operational KPIs, and executive reporting metrics.

·Proven ability to collaborate with data engineers, software engineers, product owners, compliance teams, risk teams, underwriting teams, clinical stakeholders, cloud teams, DevOps teams, and business leadership to convert complex business requirements into scalable AI/ML and GenAI solutions.

Technical Skills:

Programming Languages

Python, SQL, R, PySpark, Bash, JSON, YAML

Generative AI / LLMs

GPT-4, Claude, LLaMA, BERT, RoBERTa, T5, BioBERT, ClinicalBERT, Hugging Face Transformers, OpenAI API, Azure OpenAI, AWS Bedrock, Prompt Engineering, Prompt Optimization, RAG, Agentic AI, Tool Calling, Function Calling, Fine-Tuning, LoRA, QLoRA, PEFT, SFT, LLM Evaluation, Hallucination Detection

AI Frameworks / Orchestration

LangChain, LangGraph, LlamaIndex, Hugging Face, Sentence Transformers, spaCy, NLTK, Prompt Templates, Agent Workflows, Context Grounding, Memory Management, LLM Routing

Vector Search / RAG

FAISS, Pinecone, ChromaDB, Weaviate, Vector Search, Hybrid Search, Embeddings, Metadata Filtering, Reranking, Retrieval Optimization, Citation-Based Responses, Grounded Response Generation

Machine Learning

Scikit-learn, XGBoost, LightGBM, Logistic Regression, Random Forest, Gradient Boosting, SVM, K-Means, Classification, Regression, Clustering, Forecasting, Anomaly Detection, Fraud Detection, Risk Scoring, Churn Prediction, Recommendation Systems

Deep Learning / Computer Vision

TensorFlow, Keras, PyTorch, CNN, RNN, LSTM, Autoencoders, Transformers, ResNet, EfficientNet, OpenCV, Image Classification, Object Detection, OCR Processing, LayoutLM, Donut, Tesseract OCR

Cloud Platforms

AWS: AWS SageMaker, AWS Bedrock, Amazon S3, AWS Lambda, AWS Glue, Amazon Redshift, Amazon EC2, Amazon EKS, Step Functions, Amazon CloudWatch, AWS IAM

Azure: Azure Machine Learning, Azure OpenAI, Azure Cognitive Search / Azure AI Search, Azure Databricks, Azure Data Factory, Azure Synapse, Azure Blob Storage, Azure AKS, Azure Monitor, Azure DevOps

GCP: Vertex AI, BigQuery, Cloud Storage, Cloud Functions, Cloud Run, Cloud Monitoring

Data Engineering / Big Data

Apache Spark, PySpark, Spark SQL, Hadoop, Hive, Airflow, Kafka, ETL Pipelines, Data Lakes, Data Warehousing, Feature Engineering, Data Quality Checks, Batch Processing, Distributed Data Processing

APIs / Backend / Deployment

FastAPI, Flask, REST APIs, Microservices, Model Serving APIs, Docker, Kubernetes, AWS EKS, Azure AKS, CI/CD Pipelines, GitHub Actions, Jenkins

Databases / Storage

PostgreSQL, MySQL, Oracle, SQL Server, MongoDB, Redis, Amazon S3, Azure Blob Storage, BigQuery, Amazon Redshift

MLOps / LLMOps

MLflow, DVC, Model Registry, Model Versioning, Prompt Versioning, Model Monitoring, Drift Detection, Automated Retraining, LLM Evaluation, Prompt Regression Testing, Deployment Automation

Visualization / Governance

Power BI, Tableau, Streamlit, Plotly, Matplotlib, Seaborn, SHAP, LIME, Responsible AI, Fairness Assessment, PII Redaction, RBAC, Audit Logging, Model Cards, Data Governance

Professional Experience:

Bank of America Generative AI Engineer Charlotte, NC Aug 2024 – Present

Primary Focus: Banking GenAI, Mortgage Underwriting AI, RAG, Agentic AI, Fraud Detection, AWS MLOps

Cloud: AWS

Core Stack: Python, AWS SageMaker, AWS Bedrock, Amazon S3, AWS Lambda, Amazon EKS, LangChain, LangGraph, FAISS, GPT-4, LLaMA, FastAPI, MLflow, DVC, Docker, Kubernetes

Responsibilities:

·Designed and supported the deployment of an AI-driven mortgage underwriting platform combining document intelligence, credit risk scoring, fraud detection, and GenAI-powered policy search into a production-grade system supporting high-volume loan application workflows.

·Designed and implemented Retrieval-Augmented Generation pipelines using LangChain, FAISS, GPT-4, embeddings, semantic retrieval, prompt orchestration, and grounded response generation to help underwriters query mortgage guidelines and historical decision data using natural language.

·Built multi-agent Agentic AI workflows using LangGraph to orchestrate document extraction agents, compliance validation agents, underwriting policy agents, and risk scoring agents for end-to-end loan file processing.

·Developed LLM-powered underwriting assistants that summarized income documents, appraisals, credit reports, tax returns, policy guidelines, and supporting mortgage documents, reducing manual review effort and improving underwriting turnaround time.

·Fine-tuned LLaMA-based models using LoRA, QLoRA, and PEFT on domain-specific mortgage guidelines and underwriting policies to improve contextual accuracy for financial and compliance-related responses.

·Built LLM evaluation pipelines using ROUGE, BERTScore, retrieval relevance checks, factual grounding validation, hallucination testing, and custom business metrics to monitor response quality before production release.

·Developed document classification and OCR pipelines using LayoutLM, Donut, BERT, OCR workflows, and Python-based preprocessing to classify mortgage applications, appraisals, income verifications, tax returns, and supporting documents.

·Built ensemble credit scoring models using XGBoost, LightGBM, PyTorch, Scikit-learn, and deep neural networks to improve default prediction, underwriting risk assessment, and applicant-level risk segmentation.

·Developed fraud detection models using Isolation Forest, Autoencoders, LSTM networks, anomaly detection, and transaction/document behavior features to identify potential fraud patterns in mortgage and loan processing workflows.

·Designed feature engineering pipelines from loan applications, credit bureau attributes, borrower history, income documents, property records, fraud indicators, and underwriting decision history to improve model quality.

·Deployed real-time model-serving APIs using FastAPI, Docker, Kubernetes, AWS EKS, AWS SageMaker endpoints, and REST integration patterns to expose AI/ML predictions and GenAI outputs to downstream underwriting applications.

·Applied ONNX-based optimization, batching, caching, and inference tuning to improve model-serving performance, reduce latency, and support scalable inference for high-volume loan processing.

·Implemented MLOps workflows using MLflow, DVC, GitHub Actions, Docker, Kubernetes, AWS SageMaker, and CI/CD pipelines to support model tracking, versioning, automated deployments, and controlled release management.

·Conducted responsible AI and fairness reviews using SHAP, LIME, disparate impact analysis, bias assessment, and feature attribution to support Fair Lending, ECOA, FCRA, and internal model governance requirements.

·Implemented secure AI architecture using PII redaction, encryption, IAM controls, audit logging, content safety guardrails, secure API patterns, and role-based access controls for banking and mortgage data.

·Created executive dashboards using Power BI and Tableau to monitor risk exposure, fraud trends, underwriting KPIs, model performance, model drift, operational throughput, and Fair Lending compliance metrics.

·Collaborated with risk teams, underwriting SMEs, compliance partners, data engineers, cloud engineers, product owners, and audit stakeholders to align GenAI and ML solutions with business goals and regulatory expectations.

·Supported production deployments, prompt tuning, retrieval quality improvements, model monitoring, incident triage, performance tuning, validation reviews, and continuous improvement of banking GenAI and AI/ML platforms.

Environment: Python, SQL, Apache Spark, PySpark, Airflow, XGBoost, LightGBM, PyTorch, TensorFlow, Scikit-learn, LangChain, LangGraph, FAISS, GPT-4, LLaMA, LoRA, QLoRA, PEFT, BERT, LayoutLM, Donut, spaCy, Hugging Face Transformers, FastAPI, Docker, Kubernetes, AWS SageMaker, AWS Bedrock, Amazon S3, AWS Lambda, Amazon EKS, Amazon EC2, ONNX, MLflow, DVC, GitHub Actions, Power BI, Tableau, SHAP, LIME

Centene Corporation AI/ML Engineer / Generative AI Engineer St. Louis, MO Apr 2022 to Jul 2024

Primary Focus: Healthcare GenAI, Clinical NLP, RAG, Patient Risk Prediction, Azure ML

Cloud: Azure

Core Stack: Python, Azure Machine Learning, Azure Databricks, Azure Data Factory, Azure Cognitive Search, LangChain, LlamaIndex, GPT-4, BioBERT, ClinicalBERT, FastAPI, MLflow

Responsibilities:

·Designed and deployed GenAI-powered clinical decision support systems using LangChain, LlamaIndex, GPT-4, RAG pipelines, Azure Machine Learning, and Azure Cognitive Search to support clinical guideline search, formulary lookup, CMS policy review, and care management workflows.

·Built healthcare-focused RAG pipelines over clinical guidelines, formulary documents, policy content, EHR notes, claims records, and CMS references using document ingestion, chunking, embeddings, semantic search, metadata filtering, and grounded response generation.

·Developed LLM-powered care management assistants that helped users retrieve clinical protocols, summarize medical policy documents, and generate context-aware responses for internal healthcare workflows.

·Fine-tuned BERT, BioBERT, and ClinicalBERT models using supervised fine-tuning workflows for clinical named entity recognition, improving extraction of diagnoses, medications, procedures, and social determinants from unstructured healthcare notes.

·Built large-scale clinical NLP pipelines using spaCy, Hugging Face Transformers, BioBERT, ClinicalBERT, and Python preprocessing workflows to process EHR notes, claims records, clinical summaries, and member interactions.

·Developed patient risk stratification models using XGBoost, LSTM networks, Scikit-learn, claims features, utilization patterns, and longitudinal healthcare data to predict hospital readmission risk and support proactive care management.

·Built predictive models for chronic disease progression, ER visit forecasting, care gap identification, member risk scoring, claims trend analysis, and healthcare operational planning.

·Designed vector search workflows using Pinecone, sentence-transformer embeddings, metadata filters, and semantic ranking to retrieve relevant clinical protocols and policy content with fast response times.

·Implemented multi-turn conversational memory, safety guardrails, hallucination checks, PHI/PII masking, and response validation patterns for HIPAA-aligned healthcare chatbot workflows.

·Built automated model monitoring workflows using MLflow, Azure Data Factory, Azure Monitor, drift detection, concept drift tracking, latency monitoring, and retraining triggers to support production healthcare models.

·Created explainable AI dashboards using SHAP, Streamlit, Power BI, and model interpretation views to help clinicians understand patient risk drivers and improve adoption of AI-driven recommendations.

·Developed time-series forecasting models using ARIMA, Prophet, LSTM, and statistical modeling techniques to forecast ER visit volume, chronic disease trends, and care management workload.

·Built FastAPI-based model-serving services and batch scoring workflows to expose clinical risk predictions and NLP outputs to internal healthcare platforms.

·Authored model cards, data dictionaries, validation reports, monitoring documentation, and governance artifacts to support healthcare compliance, HIPAA requirements, and clinical review processes.

·Implemented CI/CD and MLOps practices using Docker, Kubernetes, MLflow, Azure DevOps, and automated deployment workflows to support reliable release management across environments.

·Collaborated with clinical SMEs, care managers, product teams, compliance teams, data engineers, and cloud teams to translate healthcare requirements into production-ready AI/ML and GenAI solutions.

·Supported production monitoring, data quality checks, model drift review, LLM response validation, prompt improvements, model retraining, and continuous optimization of clinical AI systems.

Environment: Python, SQL, PySpark, Azure Machine Learning, Azure AKS, Azure Databricks, Azure Data Factory, Azure Cognitive Search, Azure Synapse, LangChain, LlamaIndex, GPT-4, BioBERT, ClinicalBERT, spaCy, Hugging Face Transformers, SFT, XGBoost, LSTM, Prophet, ARIMA, Pinecone, Sentence Transformers, FastAPI, Docker, Kubernetes, MLflow, Streamlit, Power BI, SHAP, ROUGE, BERTScore

New York Life Insurance Data Scientist / ML Engineer New York,NY Dec 2019 to Mar 2022

Primary Focus: Insurance ML, Churn Prediction, Intelligent Document Processing, Recommendation Systems, AWS ML

Cloud: AWS

Core Stack: Python, AWS SageMaker, Amazon S3, AWS Lambda, Step Functions, PySpark, Scikit-learn, XGBoost, TensorFlow, LayoutLM, Tesseract OCR, MLflow

Responsibilities:

·Developed customer churn prediction models using Random Forest, XGBoost, neural network ensembles, policyholder demographics, claims history, interaction data, and customer behavior features to identify at-risk insurance customers.

·Designed intelligent document processing pipelines using Tesseract OCR, LayoutLM, CNN classifiers, Python preprocessing, and structured extraction logic to process claims documents, policy forms, and supporting customer records.

·Built NLP entity extraction modules using spaCy, Hugging Face Transformers, NLTK, and custom rules to identify claimant details, ICD-10 codes, procedure codes, coverage amounts, and key fields from claims documents.

·Developed time-series forecasting models using ARIMA, Prophet, LSTM, and statistical modeling techniques to predict product sales volume, policy demand, customer activity, and agent allocation needs.

·Created dynamic pricing and risk models using gradient boosting, survival analysis, customer lifetime value features, and risk segmentation logic to optimize premium pricing and improve pricing accuracy.

·Built recommendation engines using collaborative filtering, matrix factorization, similarity scoring, and policyholder behavior data to support cross-sell and upsell opportunities across insurance products.

·Conducted A/B testing on model-driven retention campaigns and analyzed lift, conversion, and retention impact across high-risk policyholder segments.

·Deployed models using AWS SageMaker, Step Functions, Amazon S3, Lambda, Docker, and MLflow to support scalable inference, workflow orchestration, and repeatable production deployments.

·Built data pipelines using Python, SQL, PySpark, Apache Spark, Hadoop, Airflow, and S3 data lake patterns to process claims, policy, customer, agent, and transaction datasets.

·Designed feature engineering workflows to create reusable customer-level, policy-level, claim-level, and interaction-level features for churn, pricing, recommendations, and risk analytics.

·Implemented model evaluation frameworks using ROC-AUC, precision, recall, F1-score, confusion matrix analysis, lift charts, RMSE, MAPE, and business KPI validation.

·Built monitoring and reporting workflows to track prediction quality, data quality, model performance, scoring failures, drift signals, and batch processing outcomes.

·Created dashboards using Tableau, Power BI, and Python visualization libraries to communicate churn risk, claim patterns, recommendation performance, pricing insights, and executive KPIs.

·Developed production APIs and batch scoring workflows to expose ML outputs to downstream insurance applications, campaign management platforms, and business reporting systems.

·Collaborated with actuarial teams, claims teams, product owners, marketing stakeholders, data engineers, and cloud engineers to translate insurance business needs into scalable machine learning solutions.

·Supported model validation, deployment troubleshooting, reporting automation, documentation, feature refinement, and continuous improvement of insurance AI/ML solutions.

Environment: Python, SQL, Pandas, Scikit-learn, TensorFlow, Keras, PyTorch, PySpark, Apache Spark, Hadoop, AWS SageMaker, Amazon S3, AWS Lambda, Step Functions, Amazon EC2, Docker, Airflow, Tableau, Power BI, spaCy, NLTK, Hugging Face Transformers, LayoutLM, Tesseract OCR, MLflow

Krio’s Info Solutions Pvt. Ltd. Data Scientist India May 2017 to Sep 2019

Primary Focus: NLP, Computer Vision, Predictive Analytics, ETL Automation, GCP Analytics

Cloud: GCP

Core Stack: Python, SQL, TensorFlow, Keras, Scikit-learn, GCP BigQuery, Cloud Storage, Cloud Functions, Flask, PostgreSQL, Airflow, Tableau, Power BI

Responsibilities:

·Built NLP-powered sentiment analysis engines using TF-IDF, Word2Vec, NLTK, spaCy, and Scikit-learn to analyze large customer review datasets and generate product improvement insights.

·Developed topic modeling pipelines using Gensim LDA, tokenization, text preprocessing, vectorization, and unsupervised learning to categorize customer feedback into product themes and reduce manual tagging effort.

·Built CNN-based image classification models using TensorFlow, Keras, ResNet, OpenCV, and image preprocessing workflows to support automated product quality inspection and visual classification use cases.

·Developed RNN and LSTM models for time-series anomaly detection in manufacturing sensor data, enabling predictive maintenance alerts and reducing operational downtime.

·Automated ETL pipelines using Python, SQL, Apache Airflow, Pandas, and PostgreSQL to collect, clean, transform, and load structured and unstructured datasets for analytics and ML workflows.

·Used GCP BigQuery, Cloud Storage, and Cloud Functions to support analytical querying, file processing, reporting automation, lightweight serverless workflows, and cloud-based data processing.

·Developed Flask-based REST APIs to expose analytics outputs, NLP predictions, image classification results, and anomaly detection alerts to internal applications and reporting platforms.

·Built data validation and quality checks including schema validation, duplicate checks, missing-value checks, outlier detection, reconciliation logic, and exception logging to improve pipeline reliability.

·Created feature engineering pipelines for text, image, sensor, and transactional data, improving model quality and consistency across multiple machine learning use cases.

·Developed supervised ML models using Logistic Regression, Random Forest, SVM, Gradient Boosting, and Scikit-learn pipelines for classification, regression, and customer analytics use cases.

·Built dashboards using Tableau, Power BI, Plotly, and Dash to communicate customer sentiment, product quality patterns, anomaly alerts, operational KPIs, and business insights.

·Implemented model evaluation using accuracy, precision, recall, F1-score, ROC-AUC, confusion matrix analysis, and error analysis to validate model quality before release.

·Supported deployment of analytics applications using Flask, Docker, Jenkins, PostgreSQL, AWS EC2, GCP Cloud Functions, and CI/CD workflows, improving reliability and maintainability.

·Implemented data anonymization, access controls, audit logging, and privacy-aware data handling practices to support GDPR and enterprise data governance requirements.

·Collaborated with product, QA, engineering, and business teams to gather requirements, troubleshoot issues, validate outputs, and deliver reliable data science and analytics solutions.

Environment: Python, R, SQL, Scikit-learn, TensorFlow, Keras, Word2Vec, TF-IDF, Gensim, NLTK, spaCy, Flask, GCP BigQuery, Cloud Storage, Cloud Functions, PostgreSQL, Apache Airflow, Docker, Jenkins, Tableau, Power BI, Plotly, Dash, OpenCV, ResNet, LSTM

EDUCATION

Bachelor’s degree in Computer science - Andhra University - Visakhapatnam, India

Contact this candidate