Christopher Nguyen on the qualitative and quantitative design, construction, and operation of systems with AI capabilities.
Christopher Nguyen is CEO and co-founder of Aitomatic, a startup building a platform for Industrial AI applications. Christopher previously held executive and leadership roles at organizations tasked with building machine learning solutions for traditional enterprises.
Christopher has spent the last few years leading teams tasked with building Industrial AI applications. Industrial AI involves incorporating AI capabilities into physical systems, which are often highly reliable systems that operate in challenging environments. Safety and reliability are therefore paramount in Industrial AI applications, and use cases can involve scenarios with limited or noisy training data, or limited internet connections and power supplies. To address these challenges, our conversation centered around what Christopher terms AI Engineering – an emerging discipline concerned with the qualitative and quantitative design, construction, and operation of systems with artificial-intelligence capabilities.
If you look at the history of how disciplines start, they don’t come from academia. Academia is sort of the reduction to a pattern: let’s formalize and systematize this new discipline and let’s create a syllabus and academic program around it. But typically, these new areas come from practical needs. AI Engineering is an emerging profession, some formalization of the field is needed to address all of the challenges around applying artificial intelligence technologies to the real world. Different people are coming at it from different angles. CMUs Software Engineering Institute has a program. I think that was motivated really by the relationship or the funding from the Department of Defense. So here’s the case of an organization – the DOD – that really needs a lot of quality and discipline and security around this technology of AI.
In the last four years, I’ve been part of the global AI effort Panasonic. That includes things like avionics, energy – you may know Panasonic makes all of the Tesla batteries – manufacturing, supply chain, automotive and so on. These are existing physical industries. … So these are what I call industry intuitions. That there’s something that transcends just the basic science of it, we’re starting to build systems out of these things. Let’s make sure there’s discipline around it.
Highlights in the video version:
Introduction to Christopher Nguyen, Founder and CEO of Aitomatic
What is AI Engineering?
AI is much more than ML
What led you to think about AI Engineering?
Will data practitioners be receptive to AI Engineering?
Who’s part of the AI Engineering coalition?
Look for support from online learning communities
Does AI Engineering require a degree?
Domain expertise is important and valuable
Involvement in the AI Engineering movement
Industrial Sector – key lessons
Human First AI Project
Combining Machine Learning with Domain Knowledge
Automate human knowledge assets
Domain knowledge is critical and essential
Predictive Maintenance vs. Anomaly Detection
Codify human knowledge with the right tools
What are recent trends that you find interesting?
Reliability, understandability, privacy, and security
AI Engineering can make ML less polarizing
- A video version of this conversation is available on our YouTube channel.
- Web sites for the AI Engineering consortium and the Human-First AI open source project.
- “Model Monitoring Enables Robust Machine Learning Applications”
- “Data Cascades: Why we need feedback channels throughout the machine learning lifecycle”
- “The Road to Intelligent Process Automation”
- Hamel Husain: “Deploying Machine Learning Models Safely and Systematically”
- Nikhil Muralidhar: “MLOps Anti-Patterns”
- Travis Addair: “The Future of Machine Learning Lies in Better Abstractions”
Subscribe to our Newsletter:
[Image: courtesy of Aitomatic.]