Van Gogh
Source: Wheatfield with Crows - Van Gogh

Nguyen Van Quang

AI Engineer

arrow_downward
Portrait Background

The Process

I am an AI Engineer.
I build practical AI systems across data pipelines, LLM applications, agentic RAG, and deployment-focused engineering.

Education

  • Hanoi University of Civil Engineering
    BSc in Computer Science | CPA: 3.21
    Sep 2022 - Sep 2026 (Expected)
    Awarded Scholarships for Good Academic Achievements: 2023, 2024

Languages

Python, TypeScript, SQL

ML/AI

Regression/Boosting, Neural Networks, Transformer, LLMs, AI Agents, Fine-tuning

Database

PostgreSQL, Vector databases

Data Engineering

ETL, Orchestration, Feature Engineering, Data Preprocessing

Tools

Git, Docker, Linux, AWS, Coding Agents

Spoken Languages

Vietnamese (native), English (B2)

Experience

VinSmart Future
VinSmart Future | Hanoi, Vietnam

AI Engineer Intern | May 2026 - Present

  • Participating in the process of building a high-quality data pipeline for the AI system in the tourism domain.
8SENECA
8SENECA | Hanoi, Vietnam

AI Engineer Intern | Nov 2025 - Mar 2026

Python, FastAPI, Langchain/LangGraph, OpenAI, Hugging Face, Docker, Postgres, Neo4j

  • Developed a robust AI-powered system for automating the process of creating and managing software requirements (SDLC) on Sparx Enterprise Architecture platform.
  • Implemented a custom LLM inference pipeline based on client requirements.
  • Fine-tuned a BERT model for classification and an embedding model for domain text vectorization.
  • Participated in the deployment process.

Projects

AI-native Pet E-commerce Platform

ThePawsome

github.com/quangliz/petshop

Next.js, FastAPI, PostgreSQL + pgvector, LangGraph, OpenAI API, Docker, AWS EC2/RDS

  • Built a full-stack Vietnamese pet e-commerce platform with product discovery, checkout, payment, order management, customer support, forum, and role-based admin workflows.
  • Designed and implemented Catbot, an agentic RAG assistant with planner-enforced tool calling, hybrid retrieval, pet-care knowledge grounding, pet-profile personalization, deterministic product-safety checks, streamed responses, verified product-card rendering, and human handoff.
  • Created an AI evaluation pipeline with 50 LangSmith-traced cases using LLM-as-judge scoring; achieved 96% pass rate, 97.0 average final score, 0% eval error rate, 100% required-tool/tool-order/safety/handoff accuracy, and 11.16s p50 / 22s p99 agent latency.

Horizon Awaits

Reach out for AI engineering roles, collaborations, or applied AI projects.