PhD Research Scientist Intern

Job Description

Join the team redefining how the world experiences design.

Hey, g'day, mabuhay, kia ora, 你好, hallo, vítejte!

We’re looking for current PhD students ready to bring their research into the real world and help shape the culture of AI at Canva.

Our full-time, 16 week AI Research Internship starts in September. During your internship, you’ll work directly with Canva’s AI team on a live, industry-scale project, turning part of your PhD journey into real world impact.

You’ll gain hands on experience with real data, production infrastructure and real deadlines, while learning from and working alongside the researchers and engineerings creating Canva’s next generation of AI-powered experiences.

Where and how you can work

Our global HQ is in Sydney, Australia, and as we continue to grow in our largest global market, we’ve made our way from down under to San Francisco.

Our Jackson Square campus blends the neighbourhoods vibrant downtown charm with Canva’s signature flair. The space is designed for both deep work and collaboration, along with spaces for Canvanauts to relax, recharge and connect.

This role is based in San Francisco, and we’re looking for someone who calls it home. Our hybrid way of working gives you flexibility - you’ll have the option to work from home as well as connecting and collaborating with your team in-person, on campus. We trust teams to choose the balance that empowers them to achieve their goals.

What you’d be doing in this role

As Canva scales change continues to be part of our DNA. But we like to think that's all part of the fun. So this will give you the flavour of the type of things you'll be working on when you start, but this will likely evolve.

At the moment, this role is focused on:

- Developing a synthetic data pipeline that produces conversations with our design agent — user turns, tool calls, and design-state changes — conditioned on personas and statistical criteria on requests derived from real usage.

- Combining persona-conditioned user-simulation techniques (e.g. evolved/adversarial personas, dual-control agent frameworks) with our existing intent-labeling and clustering pipeline, to keep synthetic output in-distribution with real usage.