14. ML Engineer

Evidence label: Established role family.

What they do

Develop, train, deploy, and maintain machine learning models. In the LLM era, many ML engineers have pivoted toward fine-tuning and applied AI work rather than building models from scratch. That work is increasingly concentrated at a small number of frontier labs.

Responsibilities

• Data pipeline development (collection, cleaning, transformation)
• Model training and fine-tuning
• Feature engineering
• Model serving and deployment (MLOps)
• Performance optimization and model compression
• Production monitoring for model drift

How the role is evolving

The "build a model from scratch" path is increasingly rare outside frontier labs. ML engineers in most companies now work on: fine-tuning existing models, building RAG systems, deploying and monitoring models in production, and bridging between AI engineers and data infrastructure. The practical overlap with AI engineer is large.

Required skills

Python (fluent), PyTorch or TensorFlow, distributed computing, data pipeline tools (Spark, Airflow, dbt), cloud ML platforms (SageMaker, Vertex AI, Azure ML), statistical foundations.