Start with the work, not the title
An AI job title can describe very different responsibilities. Read the role's deliverables: building model-backed applications, training models, maintaining inference systems, assessing outputs, or labeling data. Choose applications based on work you can demonstrate, rather than the presence of AI in the title.
Match your evidence to the role
For engineering roles, explain the system you built, how you evaluated it, and what happened when it failed. A useful project write-up separates model quality from latency, cost, reliability, and user experience. Research applications should explain the question, method, baselines, and limitations. Evaluation and annotation applications should demonstrate consistency and careful handling of ambiguous instructions.
Clarify the remote arrangement
Check eligible countries and states, required working hours, employment versus contracting, and any travel or office attendance. Remote does not necessarily mean location-independent. Ask which meetings require overlap and how technical decisions are documented.
Ask about evaluation and ownership
Useful interview questions include who defines success, how changes are tested, which data you may access, and who handles incidents. Ask whether you will own a production service, a prototype, a research milestone, or a queue of evaluation tasks. These expectations affect the skills and support you need.
Apply with a focused example
Use one relevant project to explain the problem, your contribution, the outcome, and what you would change. Never include confidential employer data or credentials in a portfolio. Review the original employer listing before applying, because requirements and availability can change.