Technology
Deterministic local AI systems for constrained hardware, experimental software, and small understandable machines.
Explore IntgrML Explore IntgrNN Book ConsultingWhy integer-only?
Most modern AI systems assume floating point hardware, large software stacks, and deployment environments with plenty of resources. We are interested in a different path: deterministic systems, compact models, and software that stays legible.
Integer-only computation changes the trade space. It can improve reproducibility, simplify deployment on constrained hardware, and make systems easier to reason about. It also forces discipline. The point is not novelty for its own sake. The point is to build systems that are smaller, clearer, and more portable.
IntgrML
Integer-only machine learning for reproducible deployment on edge and embedded systems.
Placeholder text about classical machine learning, deterministic inference, practical deployment, Python access, and integration into real software systems.
View RepositoryIntgrNN
Integer neural networks for compact, deterministic inference where floating-point assumptions are a liability.
Placeholder text about compact deployable models, constrained hardware, simplified deployment, and the tradeoff between raw model performance and practical control.
View RepositoryUse IntgrML when…
- You are working on classical ML or tabular problems
- You want reproducible deterministic behavior
- You want a smaller deployable system
- You need something practical rather than fashionable
Use IntgrNN when…
- You want neural network behavior in a constrained environment
- You care about compact deployment more than leaderboard results
- You want to experiment with integer-first architectures
- You need a system you can understand end to end
Technology becomes more convincing when it does something.
We are increasingly focused on demonstrations and living examples rather than libraries in isolation. The goal is to show what this technical worldview can produce in practice.
Local AI Demo
Placeholder text about a practical demonstration of local AI behavior, interactive systems, and the use of small deterministic models.
View DemoFeryl
Placeholder text about Feryl as a strange digital organism and a broader proof that small, unusual technical systems can become living software artifacts.
Meet FerylNeed help evaluating fit?
We can help you think through whether these libraries or this style of system design make sense for your project. That may mean evaluating edge deployment needs, reviewing architecture, or discussing whether a deterministic local approach is the right tradeoff.
Book Consulting