Computational infrastructure for fragmented, noisy, and probabilistic systems.
We build the infrastructure layer that turns noise into knowledge — entity resolution, graph intelligence, embedding systems, and distributed compute.
| Repo | Domain | Status |
|---|---|---|
| sandx-er | Entity Resolution — duplicate detection, record linkage, probabilistic clustering | Phase 1 |
| sandx-embed | Embedding & vector similarity — shared ANN infrastructure | Phase 1 |
| sandx-graph | Graph intelligence — knowledge graphs, neighborhood consensus | Phase 1 |
| sandx-compute | Distributed compute — GPU orchestration, consensus-aware scheduling | Phase 1 |
Raw fragmented data → [sandx-embed] → [sandx-er] → [sandx-graph]
encode resolve reason
Entity Resolution — infrastructure for resolving fragmented identities across large datasets. Applicable to healthcare records, financial databases, government data, scientific corpora, and cybersecurity systems.
Graph Intelligence — graph-based reasoning and consensus computation over resolved entity networks. Powers knowledge graphs, AI memory systems, and recommendation infrastructure.
Embedding Systems — the shared latent representation layer. Pluggable encoders, high-performance ANN indexing, cross-domain similarity. Used by all other SandX engines.
Distributed Compute — consensus-aware GPU orchestration across heterogeneous, multi-organization infrastructure. Positioned at the transition from crypto mining toward AI compute infrastructure.
SandX is deep systems infrastructure — built for engineers and researchers who process fragmented, probabilistic, real-world data at scale.
Transforming fragmented signals into coherent computational meaning.
- Phase 1 (Active) — Foundational identity: architecture, documentation, research foundations
- Phase 2 (Planned) — Open infrastructure: core engines, Python/R SDKs, reproducible benchmarks
- Phase 3 (Planned) — Platformization: APIs, managed services, commercial infrastructure