Project Gallery.
From consumer utilities that capture real behavioral signals to bespoke enterprise architectures and foundational edge AI research.
Oh That's Me (OTM)
OTM replaces biased self-quizzes with dynamic situational prompts. The deterministic BRO engine interprets behavioral signals across 72 psychological dimensions, unlocking authentic Mirror and Dipole social connections.
- 72-dimension empirical trait engine (L0–L9)
- Mirror Matching (Cosine similarity >= 0.70)
- Dipole complementarity growth discovery
- Dynamic rolling question prefetch queue
BRO Companion Engine
Decoupled into UI, orchestration, and a pure Python mathematical brain. BRO accumulates confidence without generative model drift, planned for ecosystem-wide portability across all future QMatus daily utilities.
- Decoupled 3-layer architecture (UI, Orchestrator, Deterministic Brain)
- Four relationship maturity progression stages
- Contradiction detection and confidence convergence math
- Zero-latency synchronous trait scoring
Enterprise Concurrency Architectures
Tailored to high-complexity enterprise workflows, eliminating third-party AI dependencies through self-hosted inference and strictly server-locked state persistence.
- Massive transaction concurrency with zero locking bottlenecks
- Automated business intelligence pipelines
- Server-locked Firestore security models
- Zero corporate data leakage guarantees
Edge-Optimized SLM Networks
Our research cell transitions from implementation to foundational IP ownership, training collaborative networks of domain-specific SLMs running locally without cloud egress.
- 1.8B parameter edge-optimized architectures
- Local multi-agent collaborative execution
- Near-zero energy consumption envelope
- 100% on-device private context synthesis
Have a custom architectural challenge?
We architect bespoke solutions for organizations requiring massive concurrency and localized models.