GumGum shows talent the ropes, then lets them lead the way

GumGum
GumGum
Jan 20, 2021 · 2 min read
Iris Fu is Manager, Computer Vision

Iris Fu wasn’t necessarily looking for a career in artificial intelligence when she discovered GumGum almost five years ago. She was a newly-minted chemical engineering PhD with deep experience in computational modeling of polymer peptides. “I wanted to explore new ideas,” Fu said.

By 2015, GumGum had been working with computer vision for about five years. What Fu lacked in engineering experience she made up for in curiosity and a willingness to learn. She came on board as an “image science” intern training algorithms to identify and detect logos in social media. The GumGum engineers “really took me under their wing. I basically learned everything from scratch,” she said.

Today, Fu manages the computer vision team, a team that sits within the larger engineering group. Together the CV team is working on developing a threat classification system that would allow GumGum’s AI to distinguish between, say, a soda bottle and a liquor bottle. Or between a portrait and a mug shot.

Now that computer vision is a more mature field, few engineers come to the department with as little experience as Fu initially had. But as a manager, she maintains the spirit of collaborative learning that helped her forge a new career.

With senior engineers, Fu wants to make sure they understand the business problem they’re solving for, then lets them run with the solution they think makes the most sense. With younger engineers, the goal is to educate and build confidence.

“There’s a lot of cross-team collaboration, a lot of guidance from the community that gives junior staff ideas about how to move projects forward,” Fu said. During brainstorming sessions, the CV and NLP teams come together to share what they’re working on and what they’ve learned at recent workshops or classes.

But sometimes newer talent needs a push. “They might say, ‘Oh, I don’t want to try this, it’s too risky,” said Fu. “It’s a matter of encouraging them to bring their ideas, speak up in meetings.”

That confidence is crucial, because GumGum wants its engineers to experiment and own their own projects. “Even if you spend a week chasing after something and it fails — then we know it doesn’t work. That’s great, because we can pursue other solutions,” explained Fu. And if your project succeeds, you could end up building a solution that helps the business grow.

“There’s not a lot of red tape here. If an idea works it can go into production within a few months,” said Fu. “You own it from the initial concept to delivery and everyone is here to help you make that happen.”

GumGum Culture Blog

We teach machines to see in order to solve hard problems across the media industry.