Toan Viet D.

Agent/RAG systems Expert|LLM Expert|Claude Code/Cursor Expert

ABOUT ME I'm a Kaggle competition master and a senior AI/ML engineer with over 8 years of experience. My expertise spans Agentic AI Systems, Multi-Agent Architectures, Large Language Models (LLMs), complex Agentic Retrieval-Augmented Generation (RAG) chatbots, Stable Diffusion, Computer Vision (CV), Natural Language Processing (NLP), Machine Learning, and Deep Learning. HOW I WORK I run multi-agent Claude Code and Cursor workflows: a task master coordinating 4-5 focused agents handling implementation, testing, and refactoring in parallel. High throughput without losing control. I pair this with production-ready engineering: unit tests, clear module boundaries, scalable architecture. I focus on shipping reliable systems that actually get used. EXPERTISE Agentic AI & Multi-Agent Systems I build production-grade AI agents across domains (education, tutoring, purchase advisors, real estate, and more), powered by state-of-the-art frameworks and techniques including LangChain, Agno, MCP, advanced agent memory/skills architectures, Langfuse, and LangSmith. Agentic RAG & LLM Expertise Built 50+ production RAG systems with advanced architectures such as Agentic RAG and GraphRAG (LightRAG). Core strengths: Open-source & self-hosted LLMs (vLLM, Hugging Face, llama.cpp) and closed-source providers (OpenAI, OpenRouter, Gemini, Claude). Multi-step reasoning, query expansion, dynamic chunking, hybrid search, re-ranking. Vector DBs: Qdrant, Pinecone, Chroma, PgVector, AWS OpenSearch. Data sources: CSV, Excel, MongoDB, websites, PDFs/Office files, Google Drive, Notion, Confluence, Email, Microsoft SharePoint. CloudOps Architecture Design, deploy, and self-host AI systems across AWS, GCP, Azure, SageMaker, and RunPod. Focused on scalable, stable, and cost-efficient AI infrastructure. LLMs Fine-tuning & Deployment Fine-tuned LLMs from 2B–72B parameters across domains and languages. Applied optimization techniques achieving 3–6× inference speedups. Main author of the 2k+ stars efficient LLM fine-tuning project: githubcom/stochasticai/xTuring Don't hesitate to reach out—I am committed to delivering the highest quality of work with cutting-edge AI technologies.