I asked 4 people how to break into AI, from a Microsoft PM to a Meta senior director. They shared the same, simple piece of advice.

When Patrick Leung, who joined Google in 2007, first saw a demo for a calling assistant that used AI, which would later be called Google Duplex, he was blown away by how realistic the voice sounded.

Leung joined the Duplex team in 2017. Although he'd been exposed to machine learning and AI concepts during previous company projects, he'd never worked on building models and had to retool himself on the job.

Leveraging the expertise around him, including by having lengthy conversations with colleagues who explained how the system worked, helped him become proficient in AI.

Leung witnessed Google Duplex's public launch in 2018, before leaving Google in 2019 to join a financial sciences company. He's continued to work with AI in the later stages of his career.

At the time of his interview with Business Insider, Leung said the barrier to getting into AI was lower than ever, and encouraged people to apply LLMs to real business problems. For instance, a friend with no coding experience used AI to personalize outreach messages for recruitment purposes at her job and improved her response rate.

If there aren't opportunities to do something like this in your current job, use AI in your spare time and put it on your résumé, Leung said, adding that people who can demonstrate they can wield LLMs effectively are going to find jobs.

Read Leung's full story here.

Sophia Sun essentially put Leung's advice into action when she pitched a project that would use AI to help customers at Kajabi, the creator commerce platform she worked for as a senior product manager.

The tool Sun envisioned would help content creators generate marketing content, such as blog posts and short-form videos, for TikTok and Instagram.

She started working on the project in April 2023, alongside engineering and marketing teammates. Seeing it through to its March 2024 launch was a learning curve for Sun, who had never built an AI product before.

That July, Sun started a new job at Microsoft as a senior AI product manager, and told Business Insider she thought her end-to-end experience with building an AI product helped her land the role.

Sun's playbook for breaking into AI was to find a real user problem, design a lightweight AI solution, and turn it into proof of work. Having good grades is one thing, she said, but building a product that demonstrates your abilities is another, she said.

Read Sun's story here.

When OpenAI released ChatGPT in 2022, Mostofa Adib Shakib knew the world was going to change.

He started his career in traditional software engineering at Snap Inc. and then ZipRecruiter, but became convinced he needed to build AI skills.

Shakib spent time learning about AI from books, videos, and research papers, and he built software projects to gain proficiency with agentic AI, such as a tool to help Bangladeshi professionals optimize their résumés.

In February 2025, he started an AI contractor role with Mercor, which, around the time of his interview with Business Insider, earned him a handsome sum of $6,400 a week.

Shakib decided not to go back to a full-time traditional software engineering job. He said he thought focusing on building AI skills before the market gets crowded was the right bet for him.

While being a full-time employee at a company offers stability and benefits, it would restrict his time, meaning he could only focus on gaining agentic AI skills on weekends, he said. As a full-time AI contractor, he can spread out his hours as he pleases, he added.

Shakib advised people interested in staying relevant in the tech industry to embrace change, rather than fear it, and to focus on upskilling.

Read Shakib's full story here.

Although Devi Parikh completed a Ph.D. in computer vision in 2009, before going on to become a senior director of GenAI at Meta years later, she said not to assume you need a Ph.D. to break into AI.

As the cofounder and co-CEO of the AI company Yutori, she doesn't consider Ph.D.s much when hiring, but looks for people with relevant practical experience, such as training models, she said.

Other potential avenues into interesting AI work include spending time at startups or big labs, or trying side projects that make use of open source code and online communities.

Starting and executing projects has been instrumental to her career success, Parikh said. For instance, during the COVID-19 pandemic, she started working on a YouTube series called "Humans of AI," where she interviewed AI researchers about their daily habits. This gave her more professional visibility than her research alone could.

Read Parikh's full story here.

Do you have a story to share about breaking into AI? Contact this reporter at ccheong@businessinsider.com