KDD 2025 - Applied Data Science Invited Speakers: Natalie Glance
Natalie is a lifelong learner and seasoned leader with extensive experience at startups and established companies.
Chief Engineering Officer, Duolingo
Search every verified Natalie Glance interview, podcast appearance, and on-the-record quote β each transcript cross-checked by AI and human review to confirm speaker identity. Natalie Glance, Chief Engineering Officer at Duolingo, has spoken about the company's use of generative AI to improve learning efficacy. At KDD 2025, she described a fine-tuning approach for language models that involved sampling overly difficult conversations, simplifying the responses, and tuning the model, which she said made about 70% of sessions easier for learners. She also discussed the "video call with Lily" feature, which uses AI to conduct natural conversations with learners, and noted that the team focuses on "total words spoken" as a proxy for engagement. Glance stated that the company gave the video call team a budget of millions of dollars for AI spending, allowing them to decide how to allocate funds between cost optimization and model quality. At Config 2025, Glance discussed Duolingo's engineering and design culture, noting that the company outlawed the term "MVP" in favor of "V1" to emphasize quality while pragmatically cutting corners. She said the company runs hundreds of A/B tests in parallel and described a recent change replacing the "hearts" mechanic with an "energy" system, which she said was initially met with little leadership support but proved positive for both monetization and daily active user growth. Glance also mentioned that designers are starting to use AI for "vibe coding" complex animations to better communicate their vision to engineers.
“Last year when we did it for 2025, we had this big unknown of how much we were going to spend on AI. Previously we could always predict our spend within just a few percent. For AI, it was really anybody's guess how much we would be spending in 2025. So at the end of the day, what we decided to do is we gave this team w...”
“We try to get the sweet spot there between how much we want people to optimize on cost versus using the best things so they can move as fast as possible.”
“We gave this team working on video call a budget of millions of dollars. You can spend up to this amount and you can do it by either spending a lot of time on cost optimization early in the year so you can save up money for fancier models down the line, but you figure it out. They're not super happy about this.”
“LLMs are great for education. A lot of things are great for education. Books, we've had books for a long time. They're great for education too. But it takes a lot of motivation to learn via book. What we're about as a company is making that learning process bite-sized, engaging, building a habit, making it fun. So the...”
Natalie is a lifelong learner and seasoned leader with extensive experience at startups and established companies.
Config 2025: Figma's conference for people who build products Breakthrough innovation happens when teams work togetherΒ ...
From physics researcher to software engineer practitioner? Now that's what we call a career jump! Before helping build one of theΒ ...
The fields of application of computer science are broadening every year, in very exciting ways. This talk reflects on how computerΒ ...
Natalie Glance is Director of Engineering at Duolingo, where she has been working on conversational bots for teachingΒ ...
Sign in to search the full transcript archive, filter by topic, and access every quote from Natalie Glance.