How to build an AI team: roles and hiring
Roles needed for AI products, first hires, ML engineers vs AI product engineers, interviewing and team structure.
By Editorial Team
Building AI products does not always require a big research team. Many successful AI features today are built by small product teams using hosted models well. The right roles in the right order use resources efficiently.
Core roles
- AI product engineer: a software engineer integrating hosted models via API and building prompts, RAG and evaluation infrastructure; most startups' first AI hire; see generative AI integration.
- Machine learning engineer: expert in training, fine-tuning, hosting and optimising models; needed if you build or run your own; see fine-tuning.
- Data engineer: data pipelines, quality, storage and access; see AI data strategy.
- Research scientist: when new methods must be developed; not a priority for most early startups.
- Domain expert: judges output accuracy and sets labelling rules in fields such as law, health or finance
- Product manager: owns user needs, quality criteria and prioritisation
First hires
- An engineer with strong software fundamentals who has shipped with AI tools
- Domain expertise (as an adviser if not on the founding team)
- ML and data engineers when scale or custom models require them
Interview tips
- Give a real task: design a prompt and eval set, or improve a RAG flow.
- Ask how they measure quality; see model evaluation.
- Assess awareness of cost, security and privacy; see AI security.
- Value shipping products over impressive research credentials.
For general hiring principles, see our engineering team guide.
Team structure
Embedding AI experts in product teams rather than a separate "AI department" keeps features tied to real user needs. Shared infrastructure such as evals, prompt management and cost tracking can be a common platform.
Skills across the company
Sales, support and marketing using AI tools well is also an advantage; see AI and the workforce.
Conclusion
Build the AI team step by step around product needs. For most startups the most valuable person is a product-minded engineer who connects AI to a user problem.
This guide is for general information only and is not legal, financial or investment advice. Check official sources and consult professionals for current terms.
