Hong Doan

• Built a scalable AI agent with RAG pipeline using Qdrant vector database to manage millions of embeddings, resulting in more accurate real-time financial analysis and investor profiling • Redesigned PostgreSQL/MongoDB architecture and integrated with Spring Boot backend to ensure API low latency • Implemented a cross-platform mobile app using React Native with Expo

• Developed a FastMCP server to streamline both backend API access and RAG systems via LlamaIndex for multi-agent LLMs , powering tools like Claude Desktop and AI financial assistants, while reducing deployment size • Fine-tuned DeepSeek/Mistral on remittance datasets using HuggingFace Transformers and LoRA, then benchmarked performance via real-world simulations with LangChain and FastAPI, yielding insights for production deployment • Generated synthetic datasets using open-source models with MLX LM for fine-tuning, improving response accuracy and relevance while reducing inference latency by 20% and token-level costs

• Built an admin website to manage cards and send codes via email using Next.js, Node.js, and Amazon SES, while implementing reliable payments with 2C2P • Enhanced customer shopping experience by integrating tap-and-pay SDK and PassKit, allowing customers to add gift cards to Apple/Google wallets, leading to a 25% increase in wallet adoption • Ensured code quality and streamlined deployments by applying GitHub Actions for analysis and automation, reducing deployment time by 30% and minimizing manual errors

• Handled asynchronous communication between Spring Boot microservices using Kafka, ensuring data consistency and fault tolerance • Reduced frontend teams’ workload by developing a reusable Kotlin core package for components across projects • Optimized image upload by compressing product images in the Android app and storing them in Amazon S3, reducing storage size by approximately 40% • Ensured reliable and maintainable code by applying integration and unit tests, achieving 100% function coverage, and utilizing Jenkins and SonarQube for code quality analysis

• Maintained a React-based logistics dashboard with search features powered by Elasticsearch and delivery tracking using Google Maps API, enhancing user experience with fast lookup and live location visibility • Implemented real-time updates for web and mobile apps using Node.js backend and Firebase Cloud Messaging

Built delivery system using Xamarin for mobile application and back end with ASP.NET Core