“Working in AI” covers very different jobs, from mathematical research to product strategy. Job titles are inconsistent between companies, so it helps to understand what each role typically involves. Here is a guide to the main roles and the skills they require.

Research scientist

What they do: develop new methods and models, run experiments and publish papers. In industry labs, they push the frontier of model capabilities, efficiency or safety.

Skills: deep knowledge of machine learning and mathematics, strong programming, experimental design and scientific writing. Often a PhD, though not always.

Research engineer

What they do: build the infrastructure and code that make research possible — training pipelines, large-scale experiments, evaluation frameworks — and often contribute to research directly.

Skills: excellent software engineering, distributed systems, deep learning frameworks and familiarity with research.

Machine learning engineer

What they do: turn models into reliable production systems: preparing data, training and fine-tuning models, optimizing performance, deploying and monitoring.

Skills: software engineering, machine learning fundamentals, data pipelines, cloud platforms and performance optimization.

AI engineer

A newer title, common since the rise of large language models.

What they do: build applications on top of foundation models — prompt design, retrieval-augmented generation, agents, tool integrations, evaluation and guardrails.

Skills: software engineering, APIs, understanding of model capabilities and limits, evaluation practices and product sense. Less emphasis on training models from scratch.

Data scientist

What they do: analyze data to answer business questions, build predictive models, design experiments (A/B tests) and communicate insights.

Skills: statistics, Python or R, SQL, machine learning, visualization and communication.

Data engineer

What they do: build the pipelines and platforms that collect, clean and store data — the foundation for all AI work.

Skills: SQL, distributed data processing, data warehousing and cloud infrastructure.

MLOps / ML platform engineer

What they do: create the tools and infrastructure for training, deploying, versioning and monitoring models across an organization.

Skills: DevOps, containers and orchestration, cloud platforms and ML lifecycle tools.

AI product manager

What they do: decide what AI products to build, define requirements and success metrics, coordinate engineering, design and business teams, and manage risks.

Skills: product management, enough technical understanding to judge feasibility, user research, evaluation thinking and communication.

AI safety, policy and governance roles

What they do: research alignment and safety, run red-teaming, develop policies and risk frameworks, and ensure compliance with regulation.

Skills: vary widely — from machine learning research to law, policy, ethics and security.

Domain roles using AI

Many people work with AI in roles such as computational biologist, quantitative analyst, AI-focused designer, technical writer, solutions engineer or AI trainer and evaluator. Domain expertise combined with AI literacy is increasingly valuable.

How to choose

  • Love open-ended problems and math? Research.
  • Enjoy building reliable systems? ML, AI or MLOps engineering.
  • Enjoy answering questions with data? Data science.
  • Enjoy users and strategy? Product management.
  • Care most about societal impact? Safety, policy or governance.

How roles work together

On a typical AI product team, a product manager defines the problem and success metrics; data engineers provide reliable data; ML or AI engineers build and evaluate models or model-based features; MLOps engineers keep training and deployment running; and safety, legal or governance specialists review risks. In small companies, one person may cover several of these roles, so broad skills are valuable early in a career.

What to watch next

  • Growth of AI engineer roles focused on building with foundation models.
  • Demand for evaluation and AI safety specialists.
  • Changes in entry-level hiring as AI automates routine coding.
  • Regulatory compliance roles created by laws such as the EU AI Act.

Key takeaways

  • AI careers range from research to engineering, data, product and policy.
  • ML engineers productionize models; AI engineers build apps on foundation models.
  • Data scientists answer business questions with statistics and models.
  • MLOps and data engineers build the infrastructure behind AI.
  • Domain expertise plus AI literacy is increasingly valuable.

Frequently asked questions

Do I need a PhD to work in AI?

Only for some research roles. Most engineering, data and product roles value skills and experience over degrees.

What programming language should I learn?

Python is the main language for AI and data work, along with SQL. Other languages matter for specific systems.

What is the difference between an ML engineer and an AI engineer?

ML engineers typically train and deploy models; AI engineers usually build applications using existing foundation models via APIs.