Tiberiu Rostas

- Architected and led the development of a generative AI-powered financial intelligence platform that integrated proprietary Large Language Models (LLMs) to analyze billions of transactions, delivering enhanced fraud detection capabilities and personalized consumer insights for global banking partners. - Engineered the core API services in Python using FastAPI and built a responsive front-end dashboard with React and TypeScript, culminating in a 40% reduction in model inference latency and a 25% improvement in user data interaction speeds. - Developed a sophisticated Retrieval-Augmented Generation (RAG) system utilizing LangChain, embedding models from Hugging Face Transformers, and a Pinecone vector database to provide contextual, accurate responses for a customer-facing financial chatbot service deployed on AWS. - Built and managed a highly available application infrastructure on Amazon EKS using Terraform and Helm, configuring auto-scaling node groups, VPC networking, IAM policies, and leveraging AWS Lambda for serverless post-processing tasks. - Constructed resilient, event-driven data pipelines with Apache Kafka and Kinesis to ingest real-time transaction streams, processing terabytes of data daily and storing structured aggregates in PostgreSQL and semi-structured documents in MongoDB Atlas. - Implemented a comprehensive CI/CD strategy using GitHub Actions that automated Docker image builds, static code analysis with SonarQube, and canary deployments to Kubernetes via ArgoCD, shortening the code-to-production cycle time by 60%. - Designed and maintained a reusable component library for the frontend using Storybook, managed complex application state with Redux Toolkit, and executed a rigorous testing protocol with Jest, React Testing Library, and Cypress for end-to-end validation.

- Delivered a cloud-native Artificial Intelligence platform on GCP and IBM Cloud that provided predictive analytics and resource optimization solutions for enterprise clients, fundamentally modernizing their legacy systems and enhancing operational intelligence across multi-billion dollar supply chains. - Spearheaded the backend development using Python with Flask and designed a feature-rich, dynamic frontend with Angular and NgRx, achieving a 75% reduction in data processing time for key client-side analytics dashboards. - Containerized all microservices using Docker and orchestrated their deployment on Google Kubernetes Engine (GKE), managing the application lifecycle, service mesh with Istio, and infrastructure provisioning with Pulumi and TypeScript. - Built and maintained automated CI/CD pipelines in Jenkins, integrating Maven and Webpack for build processes, JFrog Artifactory for binary management, and automated security scans with Aqua Security before deploying to production environments. - Architected a polyglot persistence strategy, engineering a scalable data layer with the distributed SQL database CockroachDB for transactional integrity and the NoSQL database Cassandra for handling high-throughput, time-series machine learning feature data. - Implemented an asynchronous task processing system using Celery workers and RabbitMQ as the message broker, decoupling long-running operations like model training and report generation to ensure API responsiveness and system resilience. - Deployed and fine-tuned machine learning models developed with TensorFlow and Scikit-learn, creating robust MLOps pipelines with Kubeflow on Kubernetes to automate the training, validation, and serving of predictive models. - Engineered a high-performance, single-page application frontend with Angular, utilizing RxJS for managing asynchronous data streams, D3.js for complex data visualizations, and maintaining a reusable component library to ensure design consistency.

- Engineered a high-volume, cloud-native digital payments gateway on Microsoft Azure for a Fortune 500 retail client, enabling secure processing of millions of transactions daily and expanding their e-commerce capabilities by integrating multiple third-party providers like Stripe and PayPal. - Developed the core transaction processing APIs using Node.js with Express.js and built the responsive merchant dashboard using React and Redux, resulting in a 30% increase in payment throughput and a 50% reduction in manual financial reconciliation efforts. - Containerized the suite of microservices with Docker and orchestrated the deployment on Azure Kubernetes Service (AKS), defining the infrastructure declaratively using ARM templates and managing release cycles with Helm charts for repeatable, reliable deployments. - Established a robust CI/CD workflow using Azure DevOps Pipelines, which automated building, testing with Jest and Mocha, and deploying applications, thereby increasing developer velocity and reducing critical production bugs by over 40%. - Designed a resilient, multi-model data architecture leveraging Azure SQL Database for relational transactional data and the NoSQL Azure Cosmos DB for storing large volumes of low-latency JSON-based payment logs and user session information. - Integrated an event-driven messaging system using Azure Service Bus to decouple the order processing, notification, and fraud detection services, ensuring high availability and fault tolerance during traffic spikes on the platform. - Secured sensitive customer and payment data to meet PCI-DSS compliance standards by managing secrets with Azure Key Vault and protecting all public endpoints with Azure API Management, which handled rate-limiting, authentication, and request throttling.

- Spearheaded the complete modernization of a client's legacy monolithic retail application by re-architecting it into a distributed, cloud-native microservices ecosystem on AWS, which enhanced system scalability and cut infrastructure overhead by 35%. - Developed the primary customer-facing single-page application using React and JavaScript (ES6+) and engineered backend APIs in Node.js (Express.js), a combination that improved user session duration by 20% and accelerated feature delivery. - Containerized all application services using Docker and managed deployments on AWS Elastic Container Service (ECS) with the Fargate serverless compute engine, provisioning all underlying resources programmatically with AWS CloudFormation. - Built and maintained a robust CI/CD pipeline in Jenkins, which orchestrated automated builds, unit testing with Jest, and seamless blue-green deployments, successfully increasing the team's release cadence from monthly to weekly. - Designed a resilient data persistence strategy by migrating the relational database to Amazon RDS for PostgreSQL for transactional integrity, while utilizing Amazon DynamoDB as a high-performance NoSQL store for product catalog and user profile data. - Engineered event-driven, asynchronous workflows by integrating AWS SQS to manage order processing queues and AWS SNS for publishing notifications to downstream systems, effectively decoupling services to prevent cascading failures. - Developed data processing and ETL scripts in Python using the Boto3 SDK, creating scheduled tasks and serverless functions with AWS Lambda that were triggered by events from Amazon S3 object storage. - Established end-to-end observability by configuring centralized logging and performance metric dashboards in AWS CloudWatch, and leveraged AWS X-Ray for distributed tracing to debug and resolve latency issues across microservices.