Start from what you already have
Most people who move into AI do not start from zero. Software engineers already know how to ship and maintain systems. Data analysts know SQL, statistics and messy data. DevOps engineers know the infrastructure models run on. Scientists in physics, biology or economics already do modelling and experiment design.
Write down what you do most in your current job, then compare it with a few AI job descriptions that interest you. The gap is usually narrower than it looks, and knowing exactly what it is tells you what to learn first.
Which AI roles suit career changers
Research scientist roles usually expect a PhD. Many other AI roles do not, and some are natural next steps from a related job.
- LLM / AI Engineer: from software engineering; building products on foundation models
- Data Scientist: from data analysis; add modelling and experiment design to SQL and statistics
- MLOps Engineer: from DevOps or platform engineering; training, serving and monitoring models
- Data Engineer: from backend or analytics engineering; the pipelines models depend on
- AI Product Manager: from product management; deciding what an AI feature should do and how to measure it
- Applied Scientist: from a quantitative PhD or research job in another field
Deep knowledge of a field such as healthcare, finance or law is valuable too. Teams building AI for those fields need people who know what a good answer looks like.
Build a portfolio that proves it
Without AI job titles on your resume, your portfolio is the evidence. It does not need to be large; it needs to show that you can take a problem from data to a measured result.
What counts as portfolio work:
One end-to-end project: data, model, evaluation and a simple deployed demo
A reproduction of a paper's results, with notes on what differed
An LLM application with a real evaluation set, not just a chat demo
Contributions to open-source ML libraries (documentation counts)
A write-up of an ML problem from your current job or field
The networking playbook
Many career changers get their first AI role through people they know, often by moving into AI work at their current employer. You do not need a large network; you need a few good conversations.
- Ask whether your current employer has an AI team, and offer to help on a project
- Find people who made the same switch and ask them one specific question
- Join ML paper reading groups, meetups or open-source communities
- Write publicly about what you are learning and building
Your 90-day action plan
A career switch feels overwhelming in the abstract. It is manageable as a 90-day sprint.
Pick one target role. Compare it with your current skills and list the gaps. Fill the biggest gap with one focused course or book. Start your portfolio project.
Finish the project, including an honest evaluation. Rewrite your resume and profile for the target role. Have three conversations with people already in it.
Apply to roles that fit, and set up job alerts for your target category. Adjust your resume based on the responses you get. Keep building and writing in public.