Advancing Taboola’s AI and Innovation with Allen Chang


A defining moment for me was a conversation with Ning Ning Yu (VP, R&D), where I learned about Taboola’s mission to support publishers and the exciting opportunities to leverage AI in our products.” Meet Allen Chang, R&D team leader, who leads comprehensive projects and a team at Taboola’s R&D center in Taiwan After starting his career as a LEAP member and data scientist at Taboola LA five years ago years, Allen jumped at the opportunity to take on a leadership role and expand his experience, where he has gained the expertise to apply data science to products, work with large-scale data, and build data science infrastructures that significantly affect Taboola and the publishers they work with.

Behind the scenes, building Taboola’s NLU solution

“When I first started working at Taboola, I started the development of the natural language understanding (NLU) solution, which was an exciting project to take on,” shared Allen.

(NLU) is a field of machine learning used to interpret and understand human language.

“At the time, Taboola had been using a third-party solution to collect NLU information, which was expensive and didn’t support many languages. During my first two years working at Taboola, our team collected data and built internal systems to replace third-party solution, saving Taboola a substantial amount of money each year. It also helped us further develop Taboola’s content taxonomy (or system) to capture even more granular information and use it for business purposes.”

Taiwan’s first two-day team building event in Yilan

From developing the NLU solution to leading the team behind it

Allen shares that one of his most exciting accomplishments (so far) has been leading the team responsible for developing Taboola’s NLU solution. Taboola NLU generates several content features that serve as the foundation for several core products, including topic insights, contextual targeting, campaign review, and recommendation systems. “This initiative has significantly impacted the business by improving our product offering and driving more targeted and effective solutions for our customers.”

Allen continues: “It has been a great achievement to gradually lead others on related projects and build company knowledge around NLU. Becoming a team lead to a group focused on maintaining and developing all NLU systems has been a great honor, especially seeing my team members grow from junior engineers to seasoned professionals. I’m especially proud to see my team members grow and succeed, achieving significant milestones along the way.”

CSR Event with KIST: Brainstorm ways to create teaching materials using GPT

Apply a growth mindset for personal and team development

Throughout his time at Taboola, Allen has adopted a continuous learning growth mindset that has helped him transform the team’s approach to NLU and other initiatives. “The best advice I’ve ever received is to always prioritize learning and stay humble and kind,” he shares, which came in handy when I encountered the occasional challenge. For example, during A/B testing for the predictive segmentation solution, the team faced difficulties due to limited experience with the A/B testing infrastructure, which caused disruption to the production system , a common challenge for data scientists.

Having dealt with similar situations, Allen focused on learning from the incident and improving the process to prevent future problems. Later he started a study group around the book Reliable Online Controlled Experiments: A Practical Guide to A/B Testing, significantly expanding the team’s understanding and helping them overcome obstacles.

Additionally, Allen notes that the weekly study group brings the team together to read and discuss insightful books such as The Effective Engineer, Clean Architecture, i Computational Advertising. He adds, “These meetings encourage continuous learning and keep our team engaged with new concepts that we can incorporate into our work.”

Departure of the ice skating team at the Aurora Ice Rink

Having an impact on Taboola

Allen goes on to say that “Every six months to a year, we dive into new initiatives in addition to NLU, such as building subscription development algorithms, building matching systems for e-commerce products, and building predictive segments,” which channel team participation. excitement about new challenges and the opportunity to help clients increase their effectiveness. From NLUs helping publishers with Topic Insights features in Taboola Newsroom to publishers being able to see their traffic across different topics, keywords and content categories and compare it to the Taboola network to see short term trends and long term, “I like to know. my work directly helps our business.”

“As I reflect on the past five years at Taboola, I’m proud of the meaningful work I’ve accomplished and the fact that I’ve been able to share my passions and creativity with the teams I’ve worked with. From having caring leaders to having the incredible opportunity to collaborate with top talent on a variety of projects, each experience has allowed me to deepen my understanding of data science. It’s this kind of work and the continued potential to innovate and contribute to a mission that really resonates with me and keeps me excited to continue growing here.”

To learn more about why we’re consistently ranked as a great place to work, check out the Taboola Life blog. Visit our races page and find out your time. #TaboolaLife



Technology