Research and patents
Research and patents
The patents and papers behind the Riskernel engine. Filed at NICE Actimize, Skyhawk Security and Outbrain; the engine’s own application was filed in November 2026.
Exhibit A · Patents
Patents, mapped to capabilities
Each group below is a field of Amir’s invention work.
Holder of 9 granted US patents (and 5 pending) in machine learning, privacy-preserving AI, and fraud detection.
The Riskernel engine: Algogen decision systems 1
Amir is the sole inventor
- The full decision lifecycle How fraud logic is proposed, checked, shared and run One application covers the Riskernel engine end to end: admitting new detection logic only after it passes checks, sharing what works across institutions while each keeps its data, real-time decisions within fixed limits, turning checked logic into fast models, and a loop that keeps improving all of it. Filed with 215 claims. US provisional 64/197,364 · patent pending
Learning across clients: federated, transfer and meta-learning 4
Amir is a named inventor
- Federated + transfer learning Anomaly detection that learns across institutions Shares learned signal between entities while each keeps its own transaction data. US 12,124,933 · granted
- Recursive federated learning Automatic fraud models for each new client Stacks each client’s model on what the consortium already learned. US 2023/0237494 · pending
- Transfer learning + clustering Fraud detection when a client has almost no fraud yet Clusters the data, picks features by their explanations, and transfers signal from elsewhere. US 2023/0267468 · pending
- Meta-learning + semi-discrete calculus A training engine that learns how to train Carries to new datasets, so new models converge faster. US 12,481,890 · granted
Fraud detection at scale 4
Amir is a named inventor
- Extreme class imbalance Supervised and unsupervised models, combined An ensemble built for client datasets where fraud is extremely rare. US 12,541,721 · granted
- Real-time scoring at scale Less memory, faster fraud decisions Keeps compact quantile summaries per customer instead of full histories. US 12,118,558 · granted
- ML-generated rules Decision rules written by machine learning Generates and validates alert rules for a fraud policy engine. US 2024/0394713 · pending
- Model testing in the cloud Testing each client’s model before it goes live Validates a tenant’s classification model inside a shared cloud service. US 12,626,151 · granted
Deep and self-supervised learning for security 4
Amir is a named inventor
- Self-supervised learning Catching attacks while they still look harmless Learns normal cloud activity without labels and flags early signs of malicious behavior. US 12,476,990 · granted
- Security posture management Cloud posture checks fed by live threat data Uses what threat detection and response sees to decide which cloud misconfigurations to fix first. US 12,634,298 · granted
- Attack-path threat assessment Scoring threats by the path an attacker is likely to take Rates risk from the routes attackers are expected to follow toward valuable assets. US 12,425,439 · granted
- Attack-path threat assessment Scoring threats by the path an attacker is likely to take Rates risk from the routes attackers are expected to follow toward valuable assets. EP 4744195 · pending
Behavioral statistics 1
Amir is a named inventor
- Signal divergence Flagging items whose clicks and real engagement disagree Compares click-through rate with downstream behavior such as time spent and conversions, then scores each title component by clicks per impression, weighted by how many independent publishers use it. US 10,607,253 · granted
Details on every patent: amirshachar.com/patents
Exhibit B · Research Monographs
Research monographs
-
Hybrid AI Architectures (AlgoGens)
Generative models propose, and deterministic checks validate and repair before anything runs. arXiv 2403.01426. See how Riskernel uses it.
-
Semi-Discrete Modeling
Amir’s research monograph on regime shifts in rare-event streams. arXiv 1012.5751.
See how this would run on your own flows.
Runs beside your current system. Nothing to rip out.