Advantech AIoT InnoWorks 2022
1st Prize — 40,000,000 VND
Recognized for an AIoT application concept using Wise-PaaS, sensor data, and applied intelligence to monitor and improve environmental conditions.
- AIoT
- Wise-PaaS
- Competition
Le Ngoc Gia HuyAI Engineer Intel
AI Engineer at Intel · since Nov 2025
Building intelligent systems with product-grade precision.
I build agentic AI systems — LLM agents that plan, call tools, and act across multi-step workflows — plus RAG applications, robotic inspection, computer vision, and AIoT, moving each one from prototype to dependable, production-minded execution.
- Agentic AI
- RAG systems
- Robotic vision
- AIoT
02Background
Calm engineering for ambitious AI products.
I care about the full path from model behavior to interfaces, infrastructure, evaluation, and user trust — designing systems reliable enough for real users, not just demos.
- 01
AI systems
RAG pipelines, prompt design, embeddings, vector databases, reranking, evaluation loops, and OpenAI model integration.
- 02
Vision & robotics
Computer vision, inspection workflows, robotic simulation, kinematics, ROS, Gazebo, and sensor-driven automation.
- 03
Founder execution
Co-founder at Code4life® — technical delivery, product thinking, community building, and practical AI education.
◆Proof & recognition
HCMUTE · Robotics and AI
Top Graduation Project Score
Awarded the highest graduation project score in the Robotics and AI program for semester 2 of 2024–2025.
- Graduation project
- Robotics & AI
- Academic
03Selected work
Systems across language, vision, robotics, and automation.
Shaped from real engineering experience — each can be expanded into a full case study with demos and source as it matures.
LLM agents & orchestration
Currently building
Agentic AI Systems
Autonomous LLM agents that plan, call tools, and act across multi-step workflows — built with orchestration graphs, memory, guardrails, and evaluation loops so agent behavior stays reliable enough for production, not just demos.
- LangGraph
- Tool-calling
- Planning
- Multi-agent
- MCP
Agentic RAG & retrieval
Agentic Document QA
A LangGraph-orchestrated agent that plans retrieval, calls search and reranking tools, and adapts across multi-step queries — over document ingestion, semantic chunking, Qdrant vector search, Supabase metadata, and OpenAI models.
- LangGraph
- Tool-calling
- Qdrant
- OpenAI API
Computer vision & robotics
AI-Based Robotic Arm Inspection
An inspection workflow combining PyTorch and TensorFlow models, OpenCV, RobotStudio simulation, MATLAB/Simulink kinematics, Inventor assembly, and real-time C# socket communication.
- PyTorch
- OpenCV
- RobotStudio
- C#
Reinforcement learning & ROS
Autonomous 3-Wheeled Vehicle
An autonomous navigation prototype built with LIDAR, Gazebo simulation, ROS, reinforcement learning, and path-planning workflows.
- ROS
- Gazebo
- LIDAR
- RL
AIoT & applied intelligence
Supermarket Air Monitoring
An AIoT concept around environmental monitoring, sensor data, and applied intelligence — recognized with first prize at Advantech AIoT InnoWorks 2022.
- AIoT
- Sensors
- Applied AI
04Agent Lab
Watch an agent think.
A hands-on replay of the agentic loop I build in production — planning, tool calls, observations, self-critique. Pick a task and run it. Scripted traces, zero API calls, everything in your browser.
01 · Pick a task
Scripted replay of real agent traces — no model calls, no data leaves this page. The production versions run on LangGraph with guardrails and trajectory-level evals.
$ task: What does our leave policy say about carrying unused days into next year?
// agent runtime ready — LangGraph-style loop
// tools registered: vector_search · rerank · web_search · notify
// pick a task, then ▸ Run
idleagent.trace
Robotics & AI, HCMUTE
Where the engineering started.
05Experience
From robotics research to applied AI engineering.
- Nov 2025 — present
AI EngineerIntelCurrent
Applied AI engineering, automation, and intelligent systems in a production-minded environment.
- 2025 — present
Co-founderCode4life®
Building learning and technology initiatives around practical coding, AI literacy, and software craft.
- Aug 2024 — Feb 2025
AI / RAG EngineerRegenX
Built a document question-answering system using RAG, Qdrant vector search, Supabase metadata, semantic chunking, reranking, and OpenAI models for internal company documents.
- Jan 2024 — Jul 2024
AI Robotics InternABB Robotics
Developed an AI-based robotic arm inspection workflow with Autodesk Inventor, MATLAB/Simulink, PyTorch, TensorFlow, OpenCV, RobotStudio, and real-time C# socket communication.
- 2024 — 2025
Robotics & AIHCMUTE
Graduation-level work recognized with the highest project score in the major for semester 2 of 2024–2025.
06Toolkit
The stack behind the work.
Practical depth across the AI lifecycle — from model behavior to edge deployment.
01
AI engineering
- RAG pipelines
- LangGraph
- Prompt design
- Embeddings
- Vector DBs
- Reranking
- Eval loops
- OpenAI API
02
Vision & robotics
- PyTorch
- TensorFlow
- OpenCV
- ROS
- Gazebo
- RobotStudio
- Kinematics
03
Data & infra
- Qdrant
- Supabase
- Python
- C#
- MATLAB/Simulink
- Socket I/O
MATLAB/Simulink04
Edge & AIoT
- TFLite
- TFLite Micro
- OpenVINO
- ONNX Runtime
- Wise-PaaS
- Sensors
07Contact
Let's build something technically serious.
Open to conversations on AI engineering, RAG applications, computer vision, robotics, AI product strategy, and founder-led technical work.
Frequently asked
Who is Lê Ngọc Gia Huy?
Lê Ngọc Gia Huy (English form: Le Ngoc Gia Huy) is an AI Engineer at Intel and co-founder of Code4life®, based in Ho Chi Minh City, Vietnam. He specializes in agentic AI systems, retrieval-augmented generation, computer vision, robotics, and AIoT.
What does Le Ngoc Gia Huy specialize in?
His core focus is agentic AI — LLM agents that plan, call tools, and act across multi-step workflows, built with LangGraph orchestration, Model Context Protocol (MCP), guardrails, and evaluation loops. That sits on top of RAG engineering (Qdrant vector search, semantic chunking, reranking) and a robotics background spanning PyTorch, OpenCV, ROS, and robotic simulation.
What recognition has Lê Ngọc Gia Huy received?
He won 1st Prize at Advantech AIoT InnoWorks 2022 (40,000,000 VND) and earned the highest graduation project score in the Robotics & AI program at HCMUTE for semester 2 of 2024–2025.
How can I contact Lê Ngọc Gia Huy?
Through LinkedIn (in/lengocgiahuy1810), GitHub (@breslee1707), or TikTok (@huyg.ai) — all linked in the contact section of this portfolio.