Learning path

Learning path: Data science, ML and AI engineering

This family is splitting in two. Classic machine learning (Machine Learning, PyTorch, scikit-learn, MLOps) remains, but LLM application engineering (LLMs, RAG, LangChain, vector databases) now matches or exceeds it in listings. Python is the common base for both.

4,022 remote positions from 975 source pagesData captured October 2026
Positions
422
in Data Science & ML
Skill-tagged
131
name at least one skill
Foundation skill
Python
14% of the family
Junior or intern
6%
of stated-level positions

1. Foundation

The most requested skills in the family; learn these first

2. Core stack

The next tier of demand; together with the foundation they cover the typical listing

3. Differentiators

The most requested skills that this family asks for at least twice as often as the market as a whole

Advice for this path

  • Python is the foundation for both branches; make it production-grade, not notebook-only.
  • For LLM roles, learn retrieval, evaluation and cost control before reaching for agent frameworks.
  • For classic ML roles, show the full lifecycle: data, training, evaluation, deployment and monitoring.
  • Cloud skills appear in the family's top skills; an ML system that only runs on a laptop is a weak portfolio piece.

Tiers are computed from the 422 Data Science & ML positions in the sample. Percentages are shares of all positions in the family, including those that name no skills.