Kavya Sai

• Architected, developed, and deployed scalable full-stack clinical trial applications using Python, FastAPI, Flask, React, TypeScript, JavaScript, REST APIs, Oracle Database, and cloud services, supporting complex clinical research workflows and high-volume transactional workloads across the complete study lifecycle.
• Designed Python-based microservices and modular backend components using FastAPI and Flask, implementing RESTful APIs, asynchronous request processing, dependency injection, middleware, exception handling, request validation, authentication, authorization, and reusable service-layer architectures.
• Developed complex React and TypeScript user interfaces for large-scale clinical trial applications, building reusable components, dynamic forms, data grids, dashboards, workflow-driven screens, modal interfaces, filtering, sorting, pagination, validation, state management, and API integrations for clinical operations users.
• Developed secure and scalable REST APIs using Python and FastAPI, integrating frontend applications with Oracle-backed services and enterprise systems while implementing OAuth/JWT-based authentication, role-based access control, input validation, API versioning, structured error handling, audit logging, and secure data transmission.
• Engineered Oracle database solutions supporting clinical trial applications, including relational data modeling, schema design, complex SQL queries, joins, stored procedures, functions, indexes, constraints, transactions, database migrations, query optimization, and performance tuning for large clinical datasets.
• Implemented Python ORM and database-access layers using SQLAlchemy and related database libraries, developing efficient repository and service patterns to manage high-volume clinical transactions while maintaining data integrity and transactional consistency.
• Developed asynchronous and event-driven Python processing components using asyncio, background workers, message queues, and scheduled processing to handle long-running clinical workflows, data synchronization, document processing, notifications, and other resource-intensive application operations without impacting UI responsiveness.
• Designed data integration and processing pipelines using Python, Pandas, NumPy, SQL, and cloud-based services to transform, validate, reconcile, and process clinical trial data originating from multiple enterprise systems and application sources.
• Designed and maintained CI/CD pipelines using Jenkins and GitHub Actions, automating source-code validation, dependency management, unit testing, integration testing, frontend builds, Python packaging, Docker image creation, security checks, deployment, configuration management, and release validation across multiple environments.
• Containerized Python microservices, APIs, background workers, and supporting application components using Docker, creating reproducible deployment environments and supporting scalable deployment across development, testing, staging, and production environments.
• Implemented automated testing using PyTest, unittest, React Testing Library, and API testing frameworks, developing unit, integration, API, regression, and end-to-end test suites to validate critical clinical workflows and maintain application quality across frequent releases.
• Applied Python development best practices including object-oriented programming, SOLID principles, design patterns, modular architecture, type hints, dependency management, logging, exception handling, configuration management, code quality standards, and reusable libraries to build maintainable enterprise applications.
• Optimized full-stack application performance by analyzing Python API execution, database queries, Oracle execution plans, frontend rendering, network requests, API payloads, and background processing, implementing caching, query optimization, indexing, pagination, lazy loading, asynchronous processing, and efficient data-access patterns.
• Developed responsive and accessible React interfaces following modern UI engineering practices and WCAG/WAI-ARIA standards, ensuring consistent functionality across browsers and devices while improving usability for clinical operations and research users.
• Implemented application-level security controls including role-based permissions, secure API endpoints, input sanitization, secrets/configuration management, session and token handling, audit trails, and controlled access to sensitive clinical and operational data.
• Integrated enterprise Oracle and third-party systems with clinical trial applications through REST APIs, database integrations, batch interfaces, and asynchronous processing mechanisms, ensuring reliable data exchange, validation, error recovery, and traceability across interconnected systems.

• Applied services knowledge to design RAG solutions that retrieve and summarize card-product policies, servicing procedures, dispute guidance, fraud controls, regulatory requirements, audit evidence, and customer-support content for faster and more consistent operational decisions.

• Designed document-ingestion pipelines with chunking, metadata, embeddings, vector indexing, retrieval, and reranking; organized content by card product, account lifecycle stage, servicing process, policy type, effective date, and regulatory category so responses remained relevant and traceable.

• Developed LangChain and OpenAI API workflows for policy question answering, dispute and case summarization, regulatory-content checks, and operational report generation, using prompt templates, structured outputs, guardrails, citations, and credit-card domain evaluation datasets.

• Built asynchronous FastAPI services that exposed GenAI and predictive capabilities to servicing, risk, compliance, and operations applications while enforcing request validation, error handling, audit-friendly logging, access controls, and reusable service components.

• Created evaluation frameworks for retrieval relevance, groundedness, completeness, latency, and hallucination risk; reviewed failures with compliance, fraud, dispute, and servicing teams and refined prompts, chunking, filters, and retrieval strategies.

• Integrated Scikit-learn, XGBoost, PyTorch, and TensorFlow models for cardholder segmentation, transaction-pattern analysis, service-demand forecasting, dispute categorization, and intelligent document classification to support customer servicing and risk operations.

• Containerized Python and AI services with Docker and supported Azure deployment, Jenkins CI/CD, configuration management, monitoring, release validation, and production troubleshooting for high-availability credit-card applications.

• Partnered with product owners, compliance specialists, fraud and risk analysts, data engineers, and application teams to convert billing, payment, APR, rewards, dispute, servicing, and regulatory rules into scalable AI solutions.

• Led design and code reviews, production-readiness assessments, and mentoring on Python, ML, RAG, prompt engineering, responsible AI, data privacy, and the safe handling of cardholder and transaction information.

• Built Python, PySpark, and SQL pipelines for consumer-banking products, combining customer, deposit-account, transaction, payment, lending, servicing, and digital-channel data for analytics and machine-learning use cases.

• Developed and maintained ML model deployment infrastructure on AWS supporting credit risk assessment and customer propensity models; implemented automated versioning for governance and traceability.
• Built CI/CD pipelines using Jenkins for automated model testing, validation, and staged deployment reducing manual overhead by 70% and enabling weekly model updates.
• Applied knowledge of customer onboarding, savings and deposit accounts, interest and balance behavior, payments, lending workflows, account servicing, and digital-channel activity when defining features, business rules, and validation criteria.
• Developed Random Forest, Logistic Regression, XGBoost, and SVM models for customer-behavior analysis, account and servicing trends, operational-demand forecasting, and identification of unusual transaction or account patterns.
• Performed feature engineering across customer profile, deposit balances, transaction frequency, payment behavior, product usage, account age, digital engagement, and servicing attributes while preventing data leakage between training, validation, and scoring datasets.
• Used hypothesis testing, ANOVA, correlation, distribution analysis, cross-validation, precision, recall, F1-score, ROC-AUC, and threshold analysis to balance model performance with consumer-banking risk and servicing requirements.
• Developed FastAPI and Flask inference services and integrated analytical outputs with downstream customer-service and operations applications, adding validation, logging, error handling, API documentation, and business-rule fallbacks.
• Optimized SQL with CTEs, window functions, joins, aggregations, indexes, and stored procedures to reconcile account and transaction data and improve the timeliness of deposit, payment, and servicing reports.
• Created Power BI dashboards for customer acquisition, deposit balances, account activity, payment and transaction trends, model outputs, and operational KPIs used by product, risk, and servicing stakeholders.
• Supported Docker packaging, Azure deployment, Jenkins pipelines, model monitoring, release validation, and production investigation of data-quality, prediction, and business-rule issues in partnership with operations, risk, data, and engineering teams.
• Implemented monitoring using Prometheus for production ML models; detected and resolved model drift issues proactively through comprehensive performance tracking.
• Developed REST APIs using FastAPI to expose ML models as production services, integrated models into business applications with comprehensive error handling.
• Optimized model inference performance achieving sub-100ms response times for real-time credit decision systems supporting high-volume transaction processing

• Administered and supported AWS production environments utilizing EC2, RDS, S3, CloudWatch, Application Load Balancer (ALB), and associated monitoring and observability solutions, ensuring application reliability, scalability, availability, and operational performance.
• Developed Python utilities using Pandas for data preprocessing, feature engineering, and model evaluation; ensured data quality and consistency across pipeline.
• Created orchestrated workflows using Python and shell scripts for scheduled model retraining and deployment; maintained infrastructure for 24/7 availability
• Developed and maintained infrastructure on AWS supporting credit risk assessment and customer propensity models; implemented automated versioning for governance and traceability.
• Built CI/CD pipelines using Jenkins for automated model testing, validation, and staged deployment reducing manual overhead by 70% and enabling weekly model updates.
• Implemented robust API layers in Python, including request and response validation, authentication, authorization, role-based access control, exception handling, middleware, logging, API versioning, and standardized error responses.
• Developed full-stack applications using Python for backend services and React, TypeScript, JavaScript, HTML, and CSS for responsive and interactive frontend experiences.
• Designed and implemented database access layers using Python, SQLAlchemy, Oracle, MySQL, and SQL, developing complex queries, stored procedures, transactions, indexing strategies, and optimized data-access patterns.