ML Engineering

ML engineering is the discipline of building, training, and optimizing machine learning models - covering data curation, model architecture design, training pipelines, and inference optimization. Its primary output is a trained model rather than an end-user application.

Details

The primary practitioners are model developers, working across the full model lifecycle: data curation, architecture selection (e.g. transformer-based LLMs), pretraining, post-training (including fine-tuning, RLHF, and alignment), and efficiency techniques like distillation and model quantization.

This distinguishes it from AI engineering, which takes trained models as a given and focuses on the application layer - prompt design, context engineering, tool orchestration, RAG pipelines, evals, and production concerns. The two disciplines overlap at boundaries like fine-tuning (where application developers may fine-tune models for specific use cases) and inference optimization, but their day-to-day toolchains and deliverables differ substantially.

Synonyms

machine learning engineering