Vu Nguyen Tuan

Designed and built production-grade AI product features using React, Next.js, TypeScript, Python, and modern LLM technologies. Developed and integrated LLM-powered workflows and AI agents with OpenAI, LangChain, and LangGraph, including structured multi-step reasoning and contextual automation. Built scalable backend services and APIs with Python/FastAPI and Node.js, integrating databases, third-party services, and AI infrastructure into end-to-end workflows. Created RAG and semantic search pipelines using embeddings and vector retrieval strategies to improve response relevance and accuracy. Contributed to system architecture, data modeling, API design, and frontend architecture with emphasis on maintainability and performance. Improved reliability and deployment workflows using Docker, cloud infrastructure, CI/CD, and production monitoring; took ownership across planning, implementation, deployment, debugging, and ongoing iteration.

Developed React and Next.js frontends with clean UI/UX and paired them with Node.js or Flask backends. Integrated third-party APIs such as Stripe, Google, and Firebase; implemented authentication, data management, and caching. Deployed applications on AWS and Firebase, ensuring uptime, scalability, and reliable CI/CD releases. Collaborated with designers and PMs to deliver user-friendly products on tight deadlines. Introduced automation scripts to improve project setup and testing efficiency by 25%.

Led design of LLM-based assistants and retrieval-augmented generation (RAG) pipelines using GPT models and vector databases. Built internal AI APIs for data search, summarization, and customer support functionality. Collaborated with cross-functional teams (backend, frontend, data) to embed AI components into existing products. Developed text analytics and semantic search tools that improved information access speed by 40%+. Managed infrastructure on AWS and Vercel for scalable API deployments with CI/CD pipelines. Contributed to internal AI frameworks standardizing prompt templates, model routing, and embedding strategies.