Danielle An is a principal engineer and generative AI architect at Meta Horizon developing tools that make interactive worlds and games easier to create. Her work combines computer graphics, filmmaking, augmented reality, digital avatars, and the practical challenges of operating AI-driven gaming platforms.
Earlier in her career, An worked at DreamWorks and Pixar on software for creating three-dimensional content. At Meta, she contributed to Instagram’s augmented-reality platform and Meta Avatars before helping architect an AI-driven mobile-gaming platform. She also served on the SIGGRAPH 2026 poster jury.
Although she built graphics tools alongside professional artists, An lacked the artistic training to produce many visual ideas herself. Her writing on creative self-expression describes AI-generated art as visual journaling and argues that generative tools can help more people turn imagination and emotion into creative work.
- Creative accessibility without sameness. Generative tools enable programmers without artistic training and artists without coding experience to build games, but easily generated prototypes often look interchangeable. An emphasizes coherent art, interface, storytelling, and gameplay, followed by repeated playtesting to identify what actual players enjoy.
- Runtime language models as game mechanics. An helps explore multiplayer games featuring AI-driven characters that act according to assigned personalities and respond unpredictably as matches unfold. These systems could also adapt cooperative play to individual abilities, helping less coordinated players participate meaningfully. Her interactive demonstration with David Hoe illustrated how unscripted characters can create experiences that differ each time they are played.
- Production reliability across agentic systems. Model behavior can affect creator prompts, nonplayer characters, content delivery, and ranking throughout a gaming platform. Model upgrades or prompt changes can destabilize existing experiences, creating difficult problems in debugging, quality assurance, inference costs, scalability, and the safety of dynamically generated content. An treats these operational constraints as fundamental design requirements for AI-generated games.
