Genefold AI — EngineeringNotes on systems, algorithms, and the engineering choices behind ArrowSpace and spectral intelligence at Genefold.https://www.genefold.ai/engineering/2026-08-27T00:00:00ZGenefold AIhttps://www.genefold.ai/Synthetic Data That Teaches: A Feedback Loop for Training Beyond Static Corporahttps://www.genefold.ai/engineering/005.html2026-08-27T00:00:00Z2026-08-27T00:00:00ZTommaso MoriondoData selection cannot create new training signal once the corpus is exhausted. This note proposes rewarding a synthetic-data generator for one thing only: whether its batches improve a learner on a fixed, held-out reference buffer.Genefold First Principles: The Curriculum Enginehttps://www.genefold.ai/engineering/004.html2026-08-24T00:00:00Z2026-08-24T00:00:00ZTommaso MoriondoOnce the pipes are fast enough, the bottleneck stops moving bytes and starts moving meaning. Measure what each data domain teaches, forecast its marginal return, and steer the sampler.Genefold First Principles: From Entropy to Epiplexityhttps://www.genefold.ai/engineering/003.html2026-08-24T00:00:00Z2026-08-24T00:00:00ZLorenzo Moriondo and Tommaso MoriondoEntropy measures uncertainty. But not all uncertainty is learnable. When compute is finite, the question shifts from how much uncertainty exists to how much structure we can extract per FLOP.Genefold First Principles: Depletion of Uncertaintyhttps://www.genefold.ai/engineering/002.html2026-07-15T00:00:00Z2026-07-15T00:00:00ZLorenzo Moriondo and Tommaso MoriondoEntropy is the irreducible uncertainty from the data source. Cross-entropy measures how close the model's distribution is to the truth; training depletes the KL divergence between the two.Designing ArrowSpace and graph wiringhttps://www.genefold.ai/engineering/001.html2026-07-08T00:00:00Z2026-07-08T00:00:00ZLorenzo MoriondoHow graph wiring in the feature-space Laplacian powers spectral indexing (SPIN) for vector retrieval, and what the experimental evidence shows on CVE and TREC-COVID.