How to be in the Top 10% of AI Engineers

With the elevation of new features in Amazon Q (as well as GitHub Copilot, Cursor, Replit, etc.) and the groundswell of warnings that AI Coding will soon do the vast majority of software development, what is an AI Engineer to think of all of these fast-moving developments? Some AI Engineers believe that they know enough, and that they are even faster moving than AI and that their jobs will not only be safe but will prosper during the time of any AI takeover of software development. After all, they believe, you need to build the foundations for AI to take over and that will be in heavy demand. Others have a different perspective and believe that while it’s true there will always be a need for many developers in the AI takeover of software development, those many developers are not the bulk of developers that are out there now and will be quite smaller in numbers than many think. So what is the difference between these two perspectives? Numbers, we can say. And sometimes to see a perspective better you not only need to see a fair comparison but on how that comparison will change over time. And that’s where we come to benchmarks. And the desire for benchmarks. And the necessity of benchmarks. Amazon Q with it’s new features in VS Code we see how AI has gained a tighter command of the development environment at different levels. What new features will see in the coming week, and months and in a year? “Amazon Q Developer comprehends your codebase context and helps complete complex tasks through natural dialog, maintaining your workflow momentum while increasing development speed.” Amazon Q Developer elevates the IDE experience with new agentic coding experience https://aws.amazon.com/blogs/aws/amazon-q-developer-elevates-the-ide-experience-with-new-agentic-coding-experience/ So now developers are beginning to see that AI is indeed competing with them and their skills and experience have to be reassessed. More and more it will become clear that the belief that you need to know what’s happening exactly is not necessary in the majority of cases. Only when building something innovative will it matter. But what is ‘something innovative’? What is the nature of building something new in an AI app when we’re not even sure how AI works? Is not every AI app innovative in some way? Isn’t the nature of intelligence to be innovative? So it’s true we have to redefine innovation in the Agentic Era. Because it’s not about creating something totally brand new anymore. A new feature can be built by AI and that’s what it does best. The power of innovation in the Agentic Age, as far as the human part of innovation, has to come through Optimizations. There is one thing that AI and it’s agents can’t keep up with: What it has never seen or hasn’t seen too much. When you bring things not seen before into coding: For instance new algorithms, new math, a new Bayesian Network, or Quantum math, anything that’s an innovation in coding itself. However, that probably won’t create a new feature. It will, however, create a new optimization. So I think that’s how the AI Engineer can compete with the so-called AI takeover of software development. They can always stay ahead through optimization. But how? The key to optimizations is monitoring. And the key to monitoring are reports and benchmarks. So in Amazon Bedrock and SageMaker it’s important to keep up with benchmarking of all kinds. For the top 10% of AI Engineers, monitoring and benchmarks will come first. That’s the key to stay ahead of AI because being a little better means being a little better than AI in a hundred different ways. Benchmarking customized models on Amazon Bedrock using LLMPerf and LiteLLM Evaluate the performance of Amazon Bedrock resources

May 6, 2025 - 16:06

How to be in the Top 10% of AI Engineers

With the elevation of new features in Amazon Q (as well as GitHub Copilot, Cursor, Replit, etc.) and the groundswell of warnings that AI Coding will soon do the vast majority of software development, what is an AI Engineer to think of all of these fast-moving developments?

Some AI Engineers believe that they know enough, and that they are even faster moving than AI and that their jobs will not only be safe but will prosper during the time of any AI takeover of software development. After all, they believe, you need to build the foundations for AI to take over and that will be in heavy demand.

Others have a different perspective and believe that while it’s true there will always be a need for many developers in the AI takeover of software development, those many developers are not the bulk of developers that are out there now and will be quite smaller in numbers than many think.

So what is the difference between these two perspectives? Numbers, we can say. And sometimes to see a perspective better you not only need to see a fair comparison but on how that comparison will change over time. And that’s where we come to benchmarks. And the desire for benchmarks. And the necessity of benchmarks.

Amazon Q with it’s new features in VS Code we see how AI has gained a tighter command of the development environment at different levels. What new features will see in the coming week, and months and in a year?

“Amazon Q Developer comprehends your codebase context and helps complete complex tasks through natural dialog, maintaining your workflow momentum while increasing development speed.”

Amazon Q Developer elevates the IDE experience with new agentic coding experience
https://aws.amazon.com/blogs/aws/amazon-q-developer-elevates-the-ide-experience-with-new-agentic-coding-experience/

So now developers are beginning to see that AI is indeed competing with them and their skills and experience have to be reassessed. More and more it will become clear that the belief that you need to know what’s happening exactly is not necessary in the majority of cases. Only when building something innovative will it matter.

But what is ‘something innovative’? What is the nature of building something new in an AI app when we’re not even sure how AI works? Is not every AI app innovative in some way? Isn’t the nature of intelligence to be innovative?

So it’s true we have to redefine innovation in the Agentic Era. Because it’s not about creating something totally brand new anymore. A new feature can be built by AI and that’s what it does best. The power of innovation in the Agentic Age, as far as the human part of innovation, has to come through Optimizations.

There is one thing that AI and it’s agents can’t keep up with: What it has never seen or hasn’t seen too much. When you bring things not seen before into coding: For instance new algorithms, new math, a new Bayesian Network, or Quantum math, anything that’s an innovation in coding itself. However, that probably won’t create a new feature. It will, however, create a new optimization.

So I think that’s how the AI Engineer can compete with the so-called AI takeover of software development. They can always stay ahead through optimization. But how?

The key to optimizations is monitoring. And the key to monitoring are reports and benchmarks. So in Amazon Bedrock and SageMaker it’s important to keep up with benchmarking of all kinds. For the top 10% of AI Engineers, monitoring and benchmarks will come first. That’s the key to stay ahead of AI because being a little better means being a little better than AI in a hundred different ways.

Benchmarking customized models on Amazon Bedrock using LLMPerf and LiteLLM

Evaluate the performance of Amazon Bedrock resources