Skip to content
VIET ANH / ENGINEERING NOTEBOOKIDEAS → SYSTEMS → LESSONS

MACHINE LEARNING LEAD / AI ARCHITECT

Building AI
beyond the demo.

I’m Viet Anh. I lead machine learning at Scopic and created AnyLabeling.

From the first architectural decision to the system people rely on — I work where hands-on engineering meets technical leadership. I build in the open and share what the real work teaches me.

Viet-Anh Nguyen outdoors among autumn trees
Viet AnhEngineer. Builder. Always learning.
A little more about me
Hands-on by choice. Curious by default.
FIG. 01 / MY WORKBENCHDRAG TO ROTATE
  1. 01
    ANYLABELING

    The labeling tool I created. Good models start with good labels.

  2. 02
    EDGE BOARD

    Where I like a model to end up: running on the device.

  3. 03
    FIELD NOTES

    What the work teaches me goes into this notebook.

  4. 04
    OPENADAS

    My driver-assistance project, built on Jetson boards.

A few things I keep coming back to.

01 / SELECTED WORK

From an idea to something useful.

All projects

OPEN SOURCE · COMPUTER VISION

AnyLabeling.

Less time labeling.
More time building.

An AI-assisted data annotation tool that brings automatic labeling and human review into one workflow. Built to help turn raw images into useful training data.
Computer visionAI-assisted annotationPython
Explore AnyLabeling 3.5k GitHub stars
AnyLabeling / annotation workspace
AnyLabeling application showing AI-assisted image annotation tools
HUMAN IN THE LOOP. AI IN THE TOOLBOX.↗

02 / TECHNICAL LEADERSHIP

The bigger picture.
The engineering detail.

My approach to leading AI work: stay close to the problem, make the tradeoffs visible, and give people a system they can understand.

Background & experience
01

Start with the real constraint.

Privacy, latency, cost, and the people using the system belong in the architecture from day one.

02

Make decisions measurable.

Benchmark the options. Document the tradeoffs. Build confidence through evidence, not a polished demo.

03

Build understanding that lasts.

Share the reasoning, write down the lessons, and make the next engineer’s work easier.

04 / FIELD NOTES

Building things. Writing it down.

All articles

Ship fast, verify fresh: how I code with AI agents across projects

How I use Orca and ppStack to work across projects and machines, and why I ask a fresh agent session to check each change, with numbers from two weeks of my own sessions. Lauren Tan’s Grok Bot talk offers another example of shipping faster while improving parts of the app.

AI Agents

October 2026 in Security: Attackers Went for the Keys, and AI Outpaced Our Patching

October 2026 showed two shifts at once: attackers went after the places where credentials pile up, and AI started finding bugs faster than most teams deploy fixes. Here is what Binance, LiteLLM, Railway, ConnectWise, PHP2Shell and Node2Shell teach, and the changes developers and IT leaders should make now.

Security

Version One Is Cheap Now. What ArtCraft's AI-Built Clones Teach Software Engineers

In two weeks, one small team used coding agents to publish open-source reimplementations of Photoshop, Premiere, Acrobat, Word and Excel. I read the repositories to see how they were built, how far along they really are, and what commercial users still depend on. My conclusion: version one of software is becoming cheap, and the valuable work moves to fit, verification, support and risk.

AI Agents

THE NEXT PAGE

Good systems start
with a conversation.

Let’s talk

Have a project, a question, or an interesting idea? I’d love to hear it.