Phuong Bouvard

Cloud Computing for Scalable Real-Time Financial Data Distribution Technical Environment: AWS, Python, FastAPI, ECS, Docker, Lambda, CDK, EventBridge, Cognito, Route53, ALB, SQL, Redshift, Redis, S3, Angular Developed and led a full-stack cloud-based platform for real-time financial data distribution capable of handling 5,000 users and delivering low-latency messaging (10 tickers/sec) via WebSocket API and Redis PubSub. Architected and implemented scalable microservices for live price calculations, using dynamic containers and sticky-session load balancing to support horizontal scaling. Integrated Azure Active Directory (AAD) with AWS Cognito to provide secure, role-based authentication and access control. Built the frontend with AgGrid, Angular, and Swift for a seamless web and mobile experience; managed deployment of the mobile app to the App Store. Implemented real-time user notifications across email, web, and mobile using AWS SNS. Managed project delivery—launching a beta within 6 months and a €500K budget, integrating smoothly with existing systems, and coordinating a 6-person team to meet stakeholder needs. Key outcome: Demonstrated a robust, scalable live price distribution service in the cloud, providing secure, optimized access and real-time financial data to thousands of users. Simplifying Database Interactions and Financial Data Analysis with ChatGPT Technical Environment: Python, AWS, FastAPI, Streamlit, SQL, LLM, Docker, ECS, CDK Built a full-stack chatbot system with MCP Agent, connecting to internal APIs for real-time answers and refinery reports. Leveraged advanced LLM to interpret user queries, extract key entities, resolve ambiguities, and translate requests into accurate SQL—making data and financial insights easily accessible for non-technical users. Power Plant Product Mapping & Trading Insights Technical Environment: AWS, Python, FastAPI, Streamlit, SQL, LLM, RAG, ECS, CDK Developed a full-stack geospatial mapping platform using AWS Textract to extract and analyze ISCC certification data, pinpointing new product deployment at power plant locations. Leveraged these insights to drive strategic long/short trading decisions Crude Arbs dashboard Technical Environment: AWS, Python, FastAPI, Streamlit, ECS, AWS CDK Developed, deployed and monitored daily data transformation pipelines and dashboard for crude arbitrary from scratch along with the infrastructure as a code with AWS CDK. Crude Blending Optimization Technical Environment: Linear Programming, Python, SQL Developed a linear programming model in Python to optimize crude oil blending for refinery operations, ensuring engineering compliance and commercial efficiency through complex constraint integration. Daily Blending Curves Construction Technical Environment: Linear Programming, Python, Timeseries, SQL Constructed daily gasoline price curves from monthly forward broker marks using constrained interpolation and time series forecasting techniques to enforce continuity and capture seasonal patterns. Incorporated seasonality effects (winter/summer) to reflect demand-driven price variations and improve short-term price prediction accuracy.

Cloud Computing for Energy Production Monitoring - Innovathon Cup Finalist, 2022 Technical Environment: AWS, Python, Swagger, Helm, ArgoCD, Kubernetes, ECS, EventBridge, AWS Batch, JFrog, ECR, GeoJSON, PostgreSQL, DMS, SCT, Redis, S3, Terraform, Agile Methodologies. Developed a full-stack cloud-based system for real-time geospatial energy monitoring, integrating geospatial and weather forecast data to enhance energy production analysis and decision-making. The scalable architecture led to adoption by 80% of the Gas and Power Trading team. Built Python microservices with Swagger APIs for seamless integration, leveraging PostgreSQL, MongoDB, S3, and Redis to optimize data storage and performance. Visualized geospatial and weather data using GeoJSON to support informed trading decisions. Migrated SQL databases to the cloud using AWS native database migration services. Orchestrated workflows with EventBridge and managed container clusters via EKS/ECS. Automated infrastructure deployment using Terraform, Helm, ArgoCD, and established CI/CD pipelines for continuous delivery. Key outcome: Delivered a robust cloud architecture that streamlined processes, automated operations, and transformed energy monitoring and trading decision-making. Live PNL
Technical Environment: AWS, Python, Swagger, HTML, CSS, Angular, Agile Methodologies. Developed a full-stack system for near real-time financial data streaming and live PNL calculation, empowering traders with up-to-date financial insights Designed and deployed a Python service for live price streaming via WebSocket, running as a Windows service and integrated with PostgreSQL for robust data management. Built a REST API for efficient data retrieval and management from PostgreSQL, and developed a frontend application using AngularJS for real-time data visualization. Established CI/CD pipelines to automate deployment and support continuous integration, ensuring rapid updates and reliable performance. Key outcome: Enabled traders to track PNL in real time, significantly improving decision-making with immediate access to live data and streamlined operational workflows. Intraday Power Price Optimiser
Technical Environment: Linear Programming, AWS, Python, FastAPI, EKS, SQL, S3, Angular Developed a optimisation tool for intraday power bidding that models supply-demand dynamics, power plant production, interconnections, and weather forecasts, delivered through an Angular web interface. Designed and integrated a comprehensive database combining supply-demand, production, interconnection, and weather data for robust simulation modeling. Deployed the model on Amazon EKS to achieve scalability and high performance, and built/maintained CI/CD pipelines for seamless EKS deployment. Developed a REST API to serve simulation results directly to the Angular frontend. Key outcome: Optimized the simulation workflow, delivering accurate, real-time intraday power price forecasts and actionable insights via an intuitive web platform.

Created and maintained data pipelines, including: - Created a service to pivot risk metrics before integrating to AWS S3 (S3, Python, MSSQL) - Created a generic service to ingest different kinds of risk data (valuation, performance) into MSSQL (Python, Swagger, MSSQL) - Added varied additional parameters to existing data ingestion services (Python, MSSQL) - Studies on AWS services (S3, IAM, Kinesis, Apache Flink, Cloudwatch, Firehorse), presented the results to the data team biweekly - Monitored production using Kibana and services logs, and provided support for users Tools: - SQL, Python, AWS, Bamboo, Git, Swagger - Testing: unit test

Executive quantitative Analysis and build predictive models for financial trading

Affiliated with King Mongkut’s University of Technology. Classification of sugarcane’s quality from their photographs using Convolutional Neural Networks. Methodologies and results: https://ieeexplore.ieee.org/document/8911950/figures#figures Reference/ Project supervisor: Dr. S. Goldin Tools and technologies: CNN, pytorch, lua, NVIDIA

Developed a RasPi robot (C), remotely controlled from an Android app and a web app (Java) Reference/ Project supervisor: Professor V. Guyot Tools: RasPi, C, Java, batch scripting, cgi, web development