AI & machine learning
Model design, training-data and serving decisions for AI teams.

Problems SpinForge models here
- Neural architecture search
- Inference request routing
- Federated-learning client selection
- Training curriculum order
- Adversarial-example search for binary networks
- Neural-network inference as an optimization
Applications
Federated Aggregator
Selects which clients join each federated-learning round.
Product page Tested prototypeCurriculum Engine
Orders learning material for the best learning sequence.
Product page Tested prototypeAdversarial Engine
Finds adversarial inputs for binary and ternary neural networks.
Product page Tested prototypeNN-to-QUBO
Runs binary-network inference as one optimization problem.
Product pageProducts with measured results
Pricing
AI & machine learning pack
$235,000 / year
$315,000 bought separately: 25% less as a pack.
- SpinForge + Atlas$90,000
- SpinRoute$75,000
- SpinCache$90,000
- SpinProof$60,000
Pilot included: Federated Aggregator, Curriculum Engine, Adversarial Engine, NN-to-QUBO (one $30,000 focused sprint, quoted with the pack).
List prices are annual licences for a named scope; written orders set the final terms. Deployment and integration use the $15,000 workload assessment or the $30,000 focused sprint.
How to read these pages
- Validated demo
- One real instance solved end to end through the application and checked against the exact answer or an explicit legality check (GPU campaign box, June 2026). Instances are small; a demo is not a benchmark.
- Tested prototype
- Working software with a passing test suite. No benchmark result is recorded yet.
- Problem type
- A problem SpinForge + Atlas can turn into a model it solves. Ask us for results on your instances.
- Products
- Catalog products with measured results on their own pages.