July 09, 2022

Article at Sydney Quantum Academy

View original

When the improbable becomes real

Maggie Liuzzi.

Maggie Liuzzi studied economics then landed a job in artificial intelligence – which led her, in a roundabout way, into the exciting world of quantum technologies.

By Wilson da Silva

WHEN MAGGIE LIUZZI, 28, completed her Bachelor of Economics in Buenos Aires, Argentina, little did she expect to end up as an engineer in artificial intelligence at Q-CTRL, a quantum tech company headquartered in Sydney that works to make a new generation of quantum technologies useful through the power of quantum control. But that’s where she is, and she’s loving it.

“Mine is an unusual path, but there are actually a lot of connections between machine learning and economics,” Maggie says. “Economics relies on a lot of maths and statistics, and that’s definitely relevant to what I do today. Just about anything in the machine learning world requires an understanding of probability.”

Her employer, Q-CTRL, has major Silicon Valley backing, and now boasts the world’s largest team of experts in quantum control engineering. Based at the Quantum Terminal (TQT), above Central Station in Sydney, the company has offices in Los Angeles and Berlin. It is led by Professor Michael Biercuk, director of the Quantum Control Lab at the University of Sydney, an inspiring physicist with a quick wit and an enviable track record for innovation and commercialisation.

But how did she go from economics to writing artificial intelligence (AI) software? It was back in Argentina while doing her economics degree that Maggie got interested in software coding, and did an internship at Google. “At the time, I thought it would just be an extra skill that differentiates me… but the more I learned about it, the more interested I became,” she recalls.

Keen to travel again after doing exchange semesters in Egypt and Canada, she applied for a job as a junior developer in Sydney after her graduation and got the role. She was so taken by coding that she also enrolled at a master’s in software development at UTS, a partner of Sydney Quantum Academy. At UTS, she was introduced to machine learning, data science and data analytics – “some of the possible paths that software gives you,” she grinned. “But it was machine learning that really grabbed me.”

She later joined the UTS Unleashed team, working with other students to create software for their Pepper social robot to compete in the international RoboCup challenge. And her team won!

That was her first real taste of AI, and it was defining. After finishing her master’s, she went on to work as an AI software engineer at autonomous vehicles LiDAR sensor company Baraja before landing the job at Q-CTRL in 2020.

Maggie competed in the international RoboCup challenge with the UTS Unleashed team — and her team won!
Maggie competed in the international RoboCup challenge with the UTS Unleashed team — and her team won!

Maggie is now applying AI techniques, such as novel machine learning algorithms, to solve problems in characterising, optimising, and automating functionality in real commercial and research-grade NISQ-era quantum computers. This requires her to guide and tweak deep reinforcement learning algorithms to wrestle with signal processing and statistics so they suppress errors in quantum hardware and deliver controlled outputs that can be used in research or industry — essentially making the available hardware sturdier and more reliable without human intervention.

“It’s really cool actually,” she says. “It can take quite a bit of effort to develop production-ready code or product features that can reliably solve machine learning problems automatically, especially at scale. I get to apply my software engineering and machine learning skills at the same time.

“Part of my role is traditional software engineering, like writing good quality code that follows best practices and uses solid source control, so the standard of the production code is high. It needs to be – that’s the code used in the end product by our customers.

“The other part is teaching AI agents to do very specific things in the quantum world, like defining certain parameters in each segment of a [control] pulse, such as amplitude or the phase. That’s important, because a lot of quantum computations are implemented using pulses. It’s a very niche, very unique use for AI.”

Maggie feels fortunate to have landed at the forefront of a burgeoning new industry. “I came to the quantum industry by luck, and I’m really enjoying it. There’s a lot of space for people to come up with new ideas and new applications,” she says.