Gorkem Yurtseven

Gorkem Yurtseven is the co-founder and chief technology officer of fal, the generative-media infrastructure company he founded with Burkay Gur in 2021. He builds the systems that turn fast-moving advances in image, video, and audio generation into reliable products for developers and enterprises.

Yurtseven studied computer engineering at the University of Pennsylvania and worked as a senior software engineer at Amazon Web Services before founding fal. The company initially provided Python cloud computing and infrastructure for data teams, attracting paying customers before its founders redirected their technical foundation toward generative media. Their account of fal’s evolution traces that shift from Python compute to specialized inference for image models, then video and audio.

Early work with Stable Diffusion expanded alongside SDXL, FLUX, and a growing range of open and proprietary models. Yurtseven describes maintaining specialized infrastructure for every new release as unsustainable: his case for model-agnostic infrastructure centers on absorbing changing architectures and workloads without rebuilding the operating foundation. Fal’s open-source developer tools extend that infrastructure to engineers deploying and consuming machine-learning models.

• Creation becomes cheaper; creativity remains essential. Generating another image or video may become inexpensive, but storytelling, artistic judgment, and creative direction still determine what merits making.

• Personalized advertising and virtual try-on. Advertising can accommodate enormous volumes of tailored, interactive content. An A24 Civil War campaign transformed participants’ selfies into toy-soldier imagery, while virtual try-on offers retailers a practical way to make online shopping more personal and visual.

• Open models compress competitive advantages. Stable Diffusion’s availability on consumer hardware showed how quickly capabilities once concentrated in major laboratories can reach independent developers. Fal also supports proprietary providers when their models broaden customers’ options.

• Real-time interactive generation. Video requires substantially more computation than still images, making latency and cost central engineering problems. Yurtseven envisions generation fast enough for scenes to respond as they unfold, opening new possibilities across games, live experiences, and storytelling.

His AI Engineer World’s Fair presentation places these applications within the progression from early computer-generated art to modern image and video models. A partnership with AWS extends fal’s specialized inference systems with cloud capacity designed to support fluctuating production demand.