Evaluated Applications builds high-performance .NET libraries. Most share one idea: computing with holographic vectors, over a runtime that tunes itself to your hardware. The tools below run those libraries entirely in your browser — no server, no upload, nothing leaves your device. Everything is on NuGet today.
the libraries, live
Paste any data — a CSV, a log, JSON lines, a spreadsheet dump — and it profiles the whole thing with a real SQL database running in your browser: what it is, what's typical, what's weird, and what's hidden inside. Then query it yourself at a live SQL prompt.
Raise a small creature that learns to navigate a world you draw. A holographic transformer for its brain, our pathfinder for its feet — it starts clueless and gets better as you watch, all on your machine.
the base everything shares
Vector-symbolic computing for .NET. Bind, bundle, and unbind hyperdimensional / phasor (HRR) vectors with a SIMD codec — the holographic primitive the rest of these libraries are built on.
dotnet add package EvaluatedApplications.PhasorAn async pipeline runtime: resource-gated concurrency with adaptive self-tuning, sagas with compensation, middleware, and a fluent builder. The shared execution engine underneath the other libraries.
Does the work of a MediatR + Polly + Dataflow stack — one dependency, ⅓ the allocation, zero dropped requests. See the benchmark →
dotnet add package EvaluatedApplications.EvalAppa holographic SQL database, embedded or networked
An embeddable holographic database: data stored as one associative store and queried three ways over
one copy — exact SQL by key, similarity by content (NEAREST), and constant-time aggregates. Real ACID
with a write-ahead log, and a durable, larger-than-memory paged engine that spills to disk.
On a 10M-invoice ledger joined to 200k customers, the dashboard rollups resolve in microseconds and stay flat as the ledger grows — same SQL, same data, single-table and cross-table joins, verified identical against DuckDB and SQL Server. See the benchmarks →
dotnet add package EvaluatedApplications.HoloDbThe .NET client for a remote HoloDb server. Connect over TLS with a token, then run SQL and bulk-load with the same surface as the in-process engine. Columnar results stay columnar on the wire.
dotnet add package EvaluatedApplications.HoloDb.ClientThe HoloDb wire contract: length-prefixed TCP framing and a columnar-preserving codec. Normally pulled in transitively by the client — you rarely reference it directly.
dotnet add package EvaluatedApplications.HoloDb.Protocola holographic transformer
A holographic transformer for .NET: HoloFormer attention with training and inference, built on the Phasor codec. Dependency-free and AOT / trim-safe, so it drops into any .NET app without native baggage.
dotnet add package EvaluatedApplications.AlgFormerOptional ILGPU / CUDA acceleration for AlgFormer: batched GPU forward + backward training and GPU inference, with automatic CPU fallback when no CUDA device is present. The core stays dependency-free.
dotnet add package EvaluatedApplications.AlgFormer.Gpumovement, tactics, and voxels
A pathfinding and game-AI SDK: navmesh pathfinding, multi-agent scheduling, dynamic obstacles, line-of-sight, fog-of-war, influence maps, and grid tactics. Headless, and parallel on the EvalApp runtime.
dotnet add package EvaluatedApplications.TracerA holographic voxel engine: level-of-detail-native spatial storage on the Phasor codec, SIMD throughout, with procedural generation. Near data stays crisp, far data degrades gracefully.
dotnet add package EvaluatedApplications.HoloVoxelThese aren't nine unrelated packages. Phasor is the vector-symbolic core — the holographic math. EvalApp is the runtime that schedules and self-tunes the heavy work. HoloDb, AlgFormer, Tracer, and HoloVoxel are what you get when you build a database, a neural network, a game-AI SDK, and a spatial engine on that same foundation.