Architecting the Era of Agentic Autonomy.
Superpose Labs builds high-performance, GPU-accelerated platforms and games. We specialize in local model orchestration, verifiable agent work, and procedural content generation.
Contact UsWhat We’re Building
Superpose Labs is an innovation studio focused on high-compute platforms. We are currently developing proprietary systems for verifiable agentic execution and GPU-accelerated media processing, using mobile gaming and procedural environments as primary demonstration architectures for our core tech.
Enhive AI
“When your own agents aren’t enough, Enhive scales the work.”
Enhive is building a marketplace for verifiable agent work. Instead of managing hundreds of agents yourself, Enhive allows teams to use private agents first, then tap into a network of independent external agents for tasks requiring scale, diversity, or true independence. From large-scale QA to research swarms, Enhive coordinates execution, verifies results, and handles payouts — making agent capacity available on demand.
WordCrawl
A mobile word-puzzle experience featuring a procedurally-aware companion. WordCrawl utilizes custom AI-driven animation logic and a dynamic backend to generate unique gameplay challenges in real-time.
Download Now →
DropTile
A video-integrated puzzle app designed for localized compute. DropTile leverages high-fidelity media processing to transform video content into interactive logical puzzles, optimized for low-latency mobile performance.
Technical Stack
Purpose-built for GPU-accelerated inference and real-time interactive applications.
Blackwell Architecture
Optimized for NVIDIA Blackwell (GB10) superchips and DGX Spark workstations. We leverage the 10 TB/s chip-to-chip interconnect and unified memory to handle massive datasets with data-center class performance on the desktop.
Edge Orchestration
Deploying NVIDIA NIMs (Inference Microservices) for low-latency, high-concurrency model execution. Our engine utilizes TensorRT-LLM to ensure deterministic, verifiable agentic responses at the edge.
Cross-Platform Core
High-performance core built with C++ and CUDA, surfaced via a custom Flutter engine. We utilize Rive for GPU-accelerated state-machine animations and dynamic, AI-driven UI transitions.
NVFP4 Optimization
Utilizing NVFP4 (4-bit) quantization to run 70B+ parameter models locally. This hierarchical scaling allows us to maintain near-BF16 accuracy while doubling throughput and reducing the memory footprint by 50%.
Spatial Architecture
Architecting 3D environments and interactive video tiles using OpenUSD (Universal Scene Description). This enables non-destructive, hierarchical scene composition and native interoperability with NVIDIA Omniverse for complex spatial computing.
Verifiable Execution
Utilizing Blackwell TEE-I/O and Confidential Computing to ensure hardware-encrypted execution. We provide verifiable agent integrity, ensuring that on-device reasoning and user data remain protected from host-level vulnerabilities.
The Lab
Based in Scottsdale, AZ, Superpose Labs is a developer-led R&D studio focused on the intersection of local AI compute and consumer software.
We are builders first — shipping real products optimized for NVIDIA RTX and TensorRT architectures. Every project we create pushes toward a future where intelligent applications run on-device: fast, private, and deeply integrated into creative workflows.
Systems architect and developer with 14+ years of experience engineering high-performance digital environments and intelligent platforms. My work focuses on the intersection of GPU-accelerated computing and local model orchestration, with a specific interest in building verifiable agentic workflows and deterministic procedural systems. At Superpose Labs, I lead the architectural design and development of proprietary engines that bring low-latency, high-compute intelligence directly to the edge.
Get in Touch
Interested in collaborating, investing, or learning more about our work? We’d love to hear from you.
Or reach us directly at labs@superposelabs.com