Our mission
AI generation should be accessible, affordable, and open.
We're building the infrastructure to make that true — for creators who generate as a craft, developers who generate at scale, and the people whose idle GPUs power it all.
Our story
Founded 2023, Singapore
In 2023, Mauvis Ledford was running diffusion inference on rented cloud GPUs for a personal creative project. The cost per batch was high enough that he started counting credits instead of experimenting freely. At the same time, gaming PCs and workstation rigs in his network were sitting idle for 16+ hours a day with no productive outlet for their VRAM. The gap between where compute existed and where it was needed was obvious.
He founded Sogni AI in Singapore that year to close that gap — a decentralized inference network that routes generation jobs to contributed GPU nodes, built around an OpenAI-compatible API so developers wouldn't need to learn new tooling to adopt it. Singapore was a deliberate choice: APAC-first routing, a regulatory environment that treats digital infrastructure companies sensibly, and a strong pool of distributed systems and ML engineering talent.
The company is angel-backed and focused on network reliability, API quality, and fair economics for both sides of the marketplace — creators who generate and node operators who contribute.
Founded 2023
Singapore, APAC-first
50+ AI models
Diffusion, video, and more
Global node network
APAC, Europe, North America
OpenAI-compatible API
Drop-in replacement
The team
People building Sogni AI
Mauvis Ledford
Founder & CEO
David Chen
Chief Technology Officer
Distributed systems and infrastructure. Previously built job scheduling systems for large-scale compute workloads.
Sara Nguyen
Head of Product
Creative tooling and B2B SaaS product background. Focuses on the creator-side experience and credit model design.
Kota Imai
Lead ML Engineer
Diffusion model fine-tuning, LoRA training, and inference optimization. Maintains the model registry and node compatibility layer.
What we believe
Our values
Open infrastructure
Generation infrastructure should be a utility, not a walled garden. We build on open standards — OpenAI-compatible API, open model registry, no proprietary SDK required. You can migrate away from Sogni AI with zero code changes. That's intentional.
Creator-first design
Features are built around how artists actually generate — iteration speed, model variety, LoRA/ControlNet support, batch jobs, and direct output download. We don't build dashboards optimized for procurement reports. The product is for people who generate thousands of images a week.
Sustainable compute
A GPU that already exists and runs idle is structurally more efficient than a new data center purpose-built to serve the same workload. Sogni's DePIN model puts existing hardware to productive use — we are not claiming carbon neutrality certificates, we are simply not building new infrastructure when idle infrastructure already exists.