From: aidotengineer
The field of AI engineering is an evolving discipline, attempting to carve out its own identity amidst established roles like machine learning (ML) engineers and software engineers [01:55:00].
Evolution of the AI Engineer Role
The state of the AI engineering industry has been a recurring theme at AI Engineer Summits. Previous summits have documented:
- The rise of the AI engineer [01:11:00].
- The identification of three types of AI engineers [01:13:00].
- The maturation and spread of the AI engineering discipline across various fields [01:19:00].
Perspectives on the AI Engineer’s Identity
There is current resistance from two sides regarding the definition of an AI engineer [01:36:00]:
- Machine Learning (ML) Engineer Point of View: From an ML perspective, an AI engineer is often seen as largely an ML engineer with the addition of “a few prompts” [01:41:00].
- Software Engineer Point of View: From a software engineering perspective, an AI engineer is considered primarily a software engineer who calls “a few LLM APIs” [01:47:00].
The speaker notes that while currently, AI engineering might be described as “90% software engineering 10% AI”, this balance is expected to shift, with AI engineering emerging as its own distinct discipline over time [01:57:00].
Linguistic Differences
Differences in language also highlight the distinct focuses of these fields:
- Machine Learning (ML) Engineers: May use terms like “test time compute” because the primary reason to run inference for them is for testing [02:27:00].
- AI Engineers: May use “inference time compute” to emphasize their focus on actual inference processes [02:30:00].
- Software Engineers: May refer to “reasoning” [02:35:00].
The Evolving Role of the AI Engineer
The job of an AI engineer is now evolving towards building agents [11:02:00]. This parallels how ML engineers build models and software engineers build software [11:04:00]. This shift indicates a specialization within the broader engineering landscape, focusing on the development and deployment of agentic AI models.