The story of preference learning in generative latent spaces
SkinDeep.ai Inc is the pioneering company behind PLGL (Preference Learning in Generative Latent Spaces) technology. Founded with the vision of making AI truly personal, we developed the first system that learns individual preferences to generate personalized content without explicit prompting.
Our breakthrough came from a simple observation: instead of asking users to describe what they want (which is often difficult), why not let them simply rate examples and learn their preferences? This insight led to PLGL - a technology that transforms how humans interact with generative AI.
Started development of preference learning technology using early generative models. Initial experiments showed promising results in learning individual aesthetic preferences.
Released the first application demonstrating PLGL technology. Users could discover their "ideal face" through simple swipe interactions. Filed provisional patent for the core methodology.
Continued research and development, exploring applications beyond facial generation. Discovered the technology's potential across multiple domains.
Decided to open source PLGL technology for community benefit. Released comprehensive documentation, code libraries, and examples to enable widespread adoption.
Steve Seguin is a serial entrepreneur with extensive experience in AI/ML, computer vision, and real-time streaming technologies. His career spans successful exits, open source leadership, and pioneering technical innovations.
Creator & Maintainer
Open-source live streaming tools with 100K+ daily active users. Rock-solid stable systems handling hundreds of terabytes of live video data monthly. Revolutionary browser-based platforms built with efficient, scalable, and secure architecture.
Chief Innovation Officer (2017 - Present), previously CTO
Pioneered live streaming technologies and the first real-time browser-based studios, enabling download-free, low-cost, and revolutionary interactive experiences long before COVID forced adoption by society. Leading innovation in ultra-low latency content delivery, real-time video mixing/switching/effects. Multiple patents pending.
Founder
Created PLGL technology - teaching AI to understand preferences through simple ratings. Patent-pending technology for preference learning in generative latent spaces. Native Android/iOS apps developed with NativeScript (code available on GitHub). Featured in app stores with YouTube demo available.
Chief Technology Officer
CTO of publicly traded visual search company (TSX:SLC). Scaled team to 100+ employees, managing both development and research teams. Built AI training infrastructure with human taggers for computer vision systems. Led development using CNNs and SVMs for major retailers.
Co-founder & CTO
Image-based product search tool and CBIR innovator. Successfully acquired in Q4 2013. Pioneered visual search on Pinterest before Pinterest itself offered it.
Why Open Source? Steve believes in the power of open innovation. By making PLGL open source, the technology becomes defensible through community adoption rather than legal barriers. This aligns with Steve's track record - VDO.Ninja's success came from being open and accessible to all.
Continuous Innovation: Steve's philosophy centers on the belief that there are always new opportunities ahead. Open sourcing PLGL ensures the core ideas remain accessible while creating space for continuous innovation. Agency and creative thought provide endless entry points to build new value.
Building for Impact: From scaling systems to handle hundreds of terabytes at VDO.Ninja, to pioneering browser-based studios at Stage TEN before society needed them, Steve has consistently built technology ahead of its time. PLGL follows this pattern - solving tomorrow's problems today.
At SkinDeep.ai Inc, we believe AI should adapt to people, not the other way around. Our mission is to make AI generation as intuitive as expressing a preference - no technical knowledge, no complex prompts, just natural interaction.
By open sourcing PLGL, we aim to democratize personalized AI generation and enable developers worldwide to build applications that truly understand individual preferences.
PLGL represents a fundamental shift in how we interact with generative AI:
Interested in PLGL technology or potential partnerships?