Sai Nikhil
Summary
Generative AI Engineer / AI/ML Engineer with 9+ years of experience designing, developing, deploying, and supporting enterprise-grade AI, Machine Learning, NLP, Generative AI, LLM, RAG, and Agentic AI solutions across banking, insurance, financial services, healthcare, and retail domains. Strong hands-on experience building LLM-powered applications, RAG pipelines, enterprise search platforms, document intelligence systems, AI assistants, semantic search solutions, and agentic AI workflows using Python, LangChain, LangGraph, LangSmith, Llamalndex, Azure OpenAI, OpenAI APIs, Hugging Face Transformers, FAISS, Pinecone, ChromaDB, and vector search frameworks. Specialized in designing end-to-end Retrieval-Augmented Generation pipelines, including document ingestion, parsing, chunking, embedding generation, vector indexing, hybrid retrieval, semantic search, prompt construction, grounded response generation, citation handling, hallucination reduction, evaluation, and production monitoring. Strong experience in Agentic AI workflows using LangGraph and LangChain, enabling tool orchestration, multi-step reasoning, workflow automation, structured outputs, function calling, tool calling, and enterprise AI task execution. Proficient in Python, SQL, PySpark, Pandas, NumPy, Scikit-learn, TensorFlow, PyTorch, XGBoost, LightGBM, BERT, RoBERTa, Hugging Face Transformers, FastAPI, Flask, MLflow, Docker, and Kubernetes. Experienced in building ML models for classification, regression, anomaly detection, fraud detection, risk scoring, churn prediction, customer segmentation, time-series forecasting, NLP, and predictive analytics. Hands-on cloud experience across Azure, AWS, and GCP, with client-specific delivery using Azure for banking GenAI and healthcare analytics, AWS for insurance MLOps and retail backend platforms, and GCP for financial ML/NLP model deployment and scalable analytics. Strong MLOps and LLMOps experience, including model versioning, prompt versioning, CI/CD pipelines, model registry, automated retraining, drift detection, response evaluation, latency monitoring, token usage tracking, observability, rollback strategy, and production support. Experienced in secure AI deployment within regulated environments, applying RBAC, PII/PHI masking, encryption, access controls, audit logging, HIPAA-aware workflows, and financial data security practices. Skilled in creating business-facing dashboards and monitoring solutions using Power BI, Tableau, Grafana, Prometheus, Azure Monitor, and Application Insights to track AI usage, model performance, feedback trends, latency, and business adoption. Collaborative engineering professional with experience working with data engineers, DevOps teams, compliance teams, business stakeholders, product teams, and enterprise architecture groups to deliver production-ready AI/ML solutions.
Overview
10
10
years of professional experience
Work History
Generative AI Engineer
Citigroup
New York, NY
05.2025 - Current
- Designed and developed enterprise Generative AI and RAG-based applications to support banking operations, internal knowledge search, compliance workflows, policy review, document summarization, and AI-assisted decision support.
- Built end-to-end Retrieval-Augmented Generation pipelines using Azure OpenAI, LangChain, Llamalndex, FAISS, Pinecone, and ChromaDB to retrieve grounded answers from internal banking documents.
- Developed document intelligence workflows for processing policies, procedures, compliance documents, operational manuals, risk reports, and internal knowledge articles.
- Implemented complete RAG lifecycle components including document parsing, chunking, metadata tagging, embedding generation, vector indexing, semantic retrieval, prompt construction, grounded response generation, and source attribution.
- Built AI-powered assistants and intelligent search capabilities to help business users ask natural-language questions across enterprise knowledge sources.
- Developed LangGraph-based agentic workflows to orchestrate retrieval, summarization, validation, tool calling, routing, and response generation across multi-step business processes.
- Created prompt engineering frameworks using structured prompts, few-shot examples, role-based prompts, prompt chaining, response constraints, and grounding rules to improve response accuracy and consistency.
- Used LangSmith for LLM tracing, prompt versioning, chain debugging, response evaluation, hallucination tracking, and production observability.
- Integrated LLM applications with enterprise APIs using FastAPI, enabling internal systems to consume summarization, semantic search, chatbot, and document intelligence capabilities.
- Orchestrated large-scale data workflows using Azure Databricks and Azure Data Factory for ingestion, transformation, feature engineering, and preparation of enterprise datasets.
- Containerized AI services using Docker and deployed them on Azure Kubernetes Service, improving scalability, reliability, and deployment consistency.
- Automated AI/ML lifecycle workflows using Azure DevOps CI/CD, supporting build, validation, deployment, rollback, and release automation.
- Managed ML lifecycle using Azure Machine Learning model registry, ensuring reproducibility, versioning, traceability, and controlled deployment.
- Built Power BI dashboards to visualize AI usage, retrieval accuracy, model performance, feedback trends, latency, and business adoption metrics.
- Implemented security controls including RBAC, encryption, access controls, audit logging, PII masking, and secure API design for banking AI workflows.
- Collaborated with compliance, cybersecurity, data engineering, DevOps, product, and business teams to deliver secure and production-ready GenAI solutions.
- Supported production monitoring, log analysis, incident triage, prompt tuning, retrieval optimization, performance tuning, and continuous improvement of deployed AI services.
- Environment
- Python, FastAPI, Azure OpenAI, LangChain, LangGraph, LangSmith, Llamalndex, FAISS, Pinecone, ChromaDB, Azure Machine Learning, Azure Databricks, Azure Data Factory, Azure Kubernetes Service, Azure Event Hubs, Azure DevOps CI/CD, Azure Monitor, Azure Application Insights, Snowflake, Power BI, Docker, Kubernetes, REST APIs, Prompt Engineering, RAG, Agentic AI, LLM Evaluation
AI/ML Engineer
Homesite Insurance
Boston, MA
01.2024 - 04.2025
- Developed AI/ML solutions for insurance analytics, including customer churn prediction, fraud detection, claims risk scoring, customer segmentation, and underwriting decision support.
- Built scalable PySpark ETL pipelines to process large insurance datasets from policy, claims, enrollment, customer interaction, and billing systems.
- Engineered feature pipelines for structured insurance data, improving signal quality for churn, fraud, and risk-scoring models.
- Developed supervised ML models using XGBoost, LightGBM, Random Forest, Logistic Regression, and SVM to support predictive analytics and business decision-making.
- Built fraud detection models to identify suspicious claims patterns using ensemble learning, anomaly detection, threshold tuning, and imbalanced classification techniques.
- Designed customer churn and retention models to help business teams identify high-risk policyholders and improve customer engagement strategies.
- Implemented NLP pipelines using Hugging Face Transformers to extract entities and classify unstructured insurance claim notes.
- Built time-series forecasting models using ARIMA and Prophet to support actuarial planning, seasonal trend analysis, and claims volume forecasting.
- Orchestrated ML training workflows using Apache Airflow and AWS SageMaker for scheduled execution, experimentation, and batch model training.
- Maintained experiment tracking, model versioning, and reproducibility using MLflow and DVC.
- Containerized ML workflows using Docker and Kubernetes, ensuring consistent execution across development, testing, and production environments.
- Evaluated models using precision, recall, ROC-AUC, F1-score, confusion matrix, lift charts, and threshold tuning to ensure strong performance on imbalanced datasets.
- Created Tableau dashboards for churn risk, fraud signals, segmentation, and model performance insights for actuarial and underwriting teams.
- Collaborated with business, actuarial, engineering, and data teams to integrate ML outputs into insurance workflows.
- Built MLOps pipelines using GitLab CI/CD, Jenkins, Docker, Kubernetes, MLflow, and AWS SageMaker to support model deployment and monitoring.
- Supported production model monitoring, retraining, defect resolution, pipeline failures, and model performance improvements.
- Environment
- Python, PySpark, SQL, R, Scikit-learn, XGBoost, LightGBM, TensorFlow, PyTorch, Hugging Face Transformers, ARIMA, Prophet, AWS SageMaker, AWS Glue, AWS S3, AWS Lambda, AWS IAM, AWS CloudWatch, Apache Airflow, MLflow, DVC, Docker, Kubernetes, GitLab CI/CD, Jenkins, Tableau
ML Engineer / Data Scientist
DTCC
Jersey City, NJ
10.2021 - 12.2023
- Built machine learning and data science solutions for financial services workflows involving risk analytics, operational insights, document classification, anomaly detection, and prediction services.
- Designed scalable ML pipelines covering data extraction, preprocessing, feature engineering, model training, validation, deployment, and monitoring.
- Developed predictive models using Scikit-learn, XGBoost, Random Forest, Logistic Regression, and ensemble techniques for classification, regression, clustering, and risk-scoring use cases.
- Performed exploratory data analysis, statistical analysis, hypothesis testing, feature importance analysis, and business trend discovery.
- Worked with large structured and semi-structured datasets using Python, SQL, Pandas, NumPy, PySpark, Spark MLlib, and BigQuery.
- Built NLP solutions for document classification, text summarization, semantic similarity, and entity extraction using spaCy, NLTK, BERT, and Hugging Face Transformers.
- Developed early-stage LLM prototypes using OpenAI APIs, LangChain, embeddings, and vector search for document Q&A and knowledge-retrieval experiments.
- Used GCP Vertex AI to manage model experimentation, training jobs, model artifacts, deployment workflows, and endpoint-based inference for selected ML use cases.
- Leveraged BigQuery and Cloud Storage to store, query, and process large-scale financial datasets used for feature engineering, model training, reporting, and analytics.
- Built API-based model serving workflows using FastAPI, Flask, Docker, and Cloud Run to expose low-latency predictions to internal enterprise applications.
- Used Cloud Functions and Pub/Sub for lightweight event-driven processing, pipeline triggers, and integration between data-processing workflows.
- Designed feature engineering workflows including encoding, scaling, feature selection, dimensionality reduction, and reusable feature pipelines.
- Implemented MLOps practices for model versioning, automated retraining, performance monitoring, endpoint monitoring, and drift detection.
- Applied Explainable AI techniques using SHAP, LIME, and feature importance methods to improve transparency and stakeholder trust.
- Created dashboards using Power BI, Tableau, Matplotlib, and Plotly to communicate model insights, trends, anomalies, and business impact.
- Collaborated with product managers, data engineers, business stakeholders, and platform teams to deliver data-driven solutions.
- Supported deployed models by monitoring failures, troubleshooting data issues, tuning model performance, and improving reliability over time.
- Environment
- Python, SQL, Pandas, NumPy, Scikit-learn, XGBoost, Random Forest, TensorFlow, PyTorch, PySpark, Spark MLlib, FastAPI, Flask, Docker, MLflow, NLP, spaCy, NLTK, BERT, Hugging Face Transformers, OpenAI APIs, LangChain, GCP Vertex AI, BigQuery, Cloud Storage, Cloud Run, Cloud Functions, Pub/Sub, Cloud Monitoring, Power BI, Tableau, Plotly, SHAP, LIME, Git
Data Scientist
Humana
Louisville, KY
03.2019 - 09.2021
- Developed healthcare analytics and machine learning solutions supporting member engagement, churn prediction, risk scoring, claims analysis, utilization insights, and operational reporting.
- Integrated healthcare claims, enrollment, pharmacy, and member interaction datasets using Azure Data Factory, Azure Data Lake, Azure Databricks, Azure Synapse, and SQL-based pipelines.
- Built datasets for analytics and ML workflows across cloud storage, reporting platforms, and downstream machine learning pipelines.
- Conducted EDA on large healthcare datasets to identify utilization trends, engagement patterns, member behavior signals, and churn indicators.
- Developed classification models using Logistic Regression, Random Forest, and XGBoost to improve healthcare risk scoring and churn prediction.
- Applied feature engineering and feature reduction techniques including PCA, correlation filtering, encoding, scaling, and cohort-based feature creation.
- Performed cohort-based time-series analysis to study seasonal healthcare usage patterns and support operational planning.
- Conducted A/B testing for engagement campaigns, measuring statistical significance and behavioral impact.
- Applied statistical techniques including hypothesis testing, confidence intervals, regression analysis, and probability distributions to support business decision-making.
- Used Azure Machine Learning for model experimentation, training workflows, model registration, and repeatable machine learning lifecycle management.
- Designed scalable data pipelines using Azure Data Lake, Azure Synapse, Azure Databricks, and Azure Data Factory for centralized storage, transformation, analytics, and reporting.
- Developed Power BI and Tableau dashboards for churn risk, healthcare utilization, engagement metrics, claims categories, and stakeholder reporting.
- Validated ML models using stratified cross-validation, precision, recall, F1-score, ROC-AUC, and cohort-level performance analysis.
- Used Azure Monitor to track pipeline execution, monitor failures, identify performance issues, and support troubleshooting for data and analytics workflows.
- Collaborated with clinical, business, compliance, and data teams to ensure data quality, governance, and accurate interpretation of analytics outputs.
- Translated data-driven insights into actionable retention strategies and business recommendations for healthcare engagement programs.
- Environment
- Python, SQL, Pandas, NumPy, Scikit-learn, XGBoost, Random Forest, Matplotlib, Seaborn, Azure Machine Learning, Azure Data Factory, Azure Data Lake, Azure Synapse Analytics, Azure Monitor, Azure SQL, Power BI, Tableau, Statistical Analysis, A/B Testing, Healthcare Claims Data
Python Developer
Aditya Birla Retail
Mumbai, India
08.2016 - 10.2018
- Developed and maintained Python-based web applications and backend services supporting retail operations, reporting workflows, inventory-related processes, and business data automation.
- Built web applications using Django and Flask, supporting backend APIs, business logic, user-facing features, and internal tools.
- Designed and implemented RESTful APIs to enable communication between frontend applications, backend services, and external systems.
- Modeled and optimized relational databases using PostgreSQL and Amazon RDS, improving query performance, indexing, and data integrity.
- Built interactive frontend components using JavaScript, jQuery, HTML, CSS, and AJAX for internal business users.
- Developed Python data processing scripts to parse JSON payloads, transform structured data, and load records into relational databases.
- Managed application storage using Amazon S3 for file uploads, static assets, and cloud-based storage workflows.
- Deployed and supported applications using AWS EC2 and AWS Elastic Beanstalk, improving hosting consistency and release reliability.
- Configured CI/CD pipelines using Jenkins and Git, automating build, testing, and deployment to AWS-hosted environments.
- Built reusable Python modules and scripts following object-oriented programming principles and clean coding practices.
- Used Python libraries such as Pandas, NumPy, Requests, SQLAlchemy, and Pydantic for data processing and backend development.
- Developed automation scripts for data extraction, file processing, ETL workflows, recurring reports, and system monitoring.
- Implemented unit testing using PyTest to improve code quality and reduce regression issues.
- Integrated dashboards using Power BI and Tableau to visualize operational metrics, reporting trends, and business KPIs.
- Optimized backend performance through SQL tuning, indexing, connection pooling, and API response improvements.
- Performed exploratory data analysis, business process planning, statistical analysis, QA coordination, and production deployment support.
- Environment
- Python, Django, Flask, JavaScript, HTML, CSS, jQuery, AJAX, JSON, REST APIs, PostgreSQL, Amazon RDS, Amazon S3, AWS EC2, AWS Elastic Beanstalk, AWS IAM, Git, Jenkins, PyTest, Pandas, NumPy, SQLAlchemy, Power BI, Tableau, Jira
Education
Bachelor's Degree - Computer Science / Engineering
Skills
- Programming Languages
- Python, SQL, PySpark, R, Bash, JavaScript, TypeScript
- Generative AI / LLM Engineering
- Generative AI, Large Language Models, RAG, Agentic AI, Prompt Engineering, Prompt Chaining, Prompt Templates, Few-Shot Prompting, Zero-Shot Prompting, ReAct Prompting, Function Calling, Tool Calling, Structured Output Generation, Context Grounding, Hallucination Reduction, LLM Evaluation, LLMOps
- LLM Frameworks / Tools
- LangChain, LangGraph, LangSmith, Llamalndex, OpenAI API, Azure OpenAI, Hugging Face Transformers, Hugging Face Datasets, FAISS, Pinecone, ChromaDB, Sentence Transformers
- AI / Machine Learning
- Scikit-learn, XGBoost, LightGBM, Random Forest, Logistic Regression, SVM, KNN, K-Means, Decision Trees, Ensemble Methods, Classification, Regression, Clustering, Feature Engineering, Model Evaluation, Anomaly Detection, Fraud Detection, Risk Scoring, Churn Prediction, Customer Segmentation
- Deep Learning / NLP
- TensorFlow, PyTorch, Keras, Neural Networks, CNN, RNN, LSTM, Transformers, BERT, RoBERTa, Text Classification, Entity Extraction, Sentiment Analysis, Summarization, Semantic Search, Document Intelligence
- Cloud Platforms
- Azure: Azure OpenAI, Azure Machine Learning, Azure Databricks, Azure Data Factory, Azure Data Lake, Azure Synapse Analytics, Azure Kubernetes Service, Azure Event Hubs, Azure DevOps, Azure Monitor, Application Insights
- AWS: AWS SageMaker, AWS Glue, Amazon S3, AWS Lambda, Amazon Redshift, Amazon Athena, Amazon RDS, AWS IAM, AWS CloudWatch, AWS EC2, AWS Elastic Beanstalk
- GCP: Vertex AI, BigQuery, Cloud Storage, Cloud Run, Cloud Functions, Pub/Sub, IAM, Cloud Monitoring
- Databases / Data Platforms
- Snowflake, SQL Server, MySQL, PostgreSQL, Amazon RDS, Amazon S3, Redshift, BigQuery
- MLOps / DevOps
- MLflow, DVC, Apache Airflow, Docker, Kubernetes, Azure DevOps CI/CD, Jenkins, GitLab CI/CD, Git, Model Registry, Model Versioning, CI/CD for ML, Automated Deployment
- API / Backend Development
- FastAPI, Flask, Django, REST APIs, Microservices, API Integration, JSON, Swagger, Authentication, Authorization
- Visualization / Monitoring
- Power BI, Tableau, Grafana, Prometheus, Matplotlib, Azure Monitor, Application Insights, AWS CloudWatch, GCP Cloud Monitoring
- Security / Compliance
- HIPAA, RBAC, PII/PHI Masking, Encryption, Secure Data Handling, IAM, Access Controls, Audit Logging
- Methodologies
- Agile Scrum, Kanban, SDLC, Cross-Functional Collaboration, Production Support, Code Reviews, Technical Documentation
Timeline
Generative AI Engineer
Citigroup
05.2025 - Current
AI/ML Engineer
Homesite Insurance
01.2024 - 04.2025
ML Engineer / Data Scientist
DTCC
10.2021 - 12.2023
Data Scientist
Humana
03.2019 - 09.2021
Python Developer
Aditya Birla Retail
08.2016 - 10.2018