About & Credentials

Who is Amir Shachar?

Fraud Scientist, Patented Inventor, and Founder & Chief Scientist of Riskernel

Last updated: Primary sources linked on /facts/ and /patents/

Short Answer

Amir Shachar is a fraud scientist, patented inventor, and founder and Chief Scientist of Riskernel. Previously Chief AI Scientist at Skyhawk Security and Head of Fraud Data Science at NICE Actimize (2019 CEO Excellence Award), Amir Shachar is named inventor on 14 patents (9 granted US, 5 pending) and authored the Algogens framework and Semi-discrete Calculus.

Amir Shachar, fraud scientist and founder of Riskernel

Key Verified Distinctions

  • Patented Federated Learning at Scale: Led fraud data science at NICE Actimize and earned the 2019 CEO Excellence Award for commercial federated learning across Tier-1 banks.
  • 14 Named-Inventor Patents (9 Granted US, 5 Pending): Covers extreme class imbalance, federated sharing, quantile estimation, semi-discrete meta-learning, cloud security, and Algogens rules.
  • Propose-Then-Verify AI Architecture (Algogens): Introduced in the 560-page monograph arXiv:2403.01426, enforcing deterministic verification before generative rules or actions run.
  • Evaluation & Model-Risk Rigor: Focuses on fixed analyst alert budgets, delayed and censored fraud labels, and verifiable system specifications under strict NDA discipline.

1. Amir Shachar In One Line

This summary table pairs Amir Shachar's canonical public identity, roles, intellectual property, and academic record with their primary reference pages.

Attribute Verified Detail Primary Source
Canonical Identity Amir Shachar, fraud scientist amirshachar.com/#person
Current Role Founder and Chief Scientist, Riskernel (live payment and payout fraud scoring engine and advisory lab) riskernel.com/about/
Prior Industry Leadership Chief AI Scientist at Skyhawk Security; Head of Fraud Data Science at NICE Actimize (2019 CEO Excellence Award) /facts/#overview
Patents Portfolio 14 patents (9 granted US, 5 pending; 2014 to 2026) across NICE Ltd., Skyhawk Security, Outbrain, and Riskernel /patents/
Academic Training & Teaching Mathematics, Computer Science, and Statistics at Technion and Tel Aviv University; Technion TA across 4 courses (2008 to 2010) /teaching/
Research Frameworks & Corpus Algogens (arXiv:2403.01426, 560 pp.), Semi-discrete Calculus (arXiv:1012.5751), Inquiry Policy Learning, 808-entry Detachment Library /publications/
Verification & NDA Policy Every published figure traces to USPTO/Google Patents filings, arXiv preprints, or committed repo data; client NDA data omitted /facts/

Last updated: . Provenance note: Every entry above derives directly from public patent filings, arXiv metadata, Riskernel role disclosures, and committed datasets on amirshachar.com.

2. Who has Amir Shachar worked with across fraud AI, security, and academia?

Chronological record from university teaching and mathematical calculus preprints to Tier-1 banking fraud, cloud AI security, and Riskernel.

  1. 2024 to Present

    Founder and Chief Scientist, Riskernel

    Founded Riskernel to apply propose-then-verify Algogens validation and cross-institution federated learning to real-time payment and payout fraud scoring. Sole inventor on US provisional patent 64/197,364 (2026) and author of the four canonical fraud data science essays hosted on riskernel.com.

  2. 2022 to 2024

    Chief AI Scientist, Skyhawk Security

    Directed AI-driven cloud threat detection research and co-led Skyhawk Security's AI purple team. Named co-inventor on 3 granted US patents (US 12,634,298 B2; US 12,476,990 B2; US 12,425,439 B2) covering self-supervised threat detection, posture management, and anticipated attacker pathways; quoted in SC Media, TechTarget, CSO, Dark Reading, startuphub.ai, and Yahoo Finance.

  3. NICE Actimize Tenure

    Head of Fraud Data Science, NICE Actimize

    Led R&D and deployed machine learning fraud models across global Tier-1 financial institutions. Received the 2019 NICE Actimize CEO Excellence Award for patented commercial deployment of federated learning in fraud detection; named inventor on 9 NICE Ltd. patents (5 granted US, 4 pending) spanning extreme class imbalance, quantile summaries, model testing, and semi-discrete meta-learning.

  4. 2014

    Outbrain Inc. Patent Co-Inventor

    Named co-inventor on granted US Patent 10,607,253 B1 ("Content Title User Engagement Optimization"), assigned to Outbrain Inc.

  5. 2008 to 2010

    Technion Teaching Assistant & University Training

    Studied Mathematics, Computer Science, and Statistics at the Technion - Israel Institute of Technology and Tel Aviv University. Served as Teaching Assistant at the Technion for Calculus 1 (2008), Advanced Calculus (2008 & 2009), Linear Algebra 1 (2009), and Image Processing (2010), alongside early arXiv preprints on Integral Image Calculus (1005.1418) and Semi-discrete Calculus (1012.5751).

3. How does Amir Shachar approach fraud AI and model risk in practice?

Four field insights drawn from deploying banking fraud systems, designing cloud threat detectors, and auditing high-stakes AI workflows.

Alert budgets beat full-curve AUC in live banking fraud

In production fraud operations, review queues cap daily alerts at fixed analyst capacity. Models optimized on global ROC-AUC often waste top-decile precision on stale or unactionable patterns rather than maximizing confirmed loss prevented inside the alert cap.

Delayed and approval-censored labels distort naive offline validation

Chargebacks arrive weeks after transaction approval, and declined transactions never produce ground-truth fraud labels. Time-aware splits and censoring-aware evaluation prevent deceptive offline validation scores.

Generative models must propose inside deterministic validators

Because unconstrained LLM agents struggle with exact audit replay and irreversible payment risk, Algogens separates candidate rule/feature generation from deterministic checks for schema integrity, grounding, policy bounds, and blast radius.

Every headline number requires explicit provenance and failure boundaries

Across 808 Detachment Library cards and four fraud essays, documenting what an authority or model cannot see matters as much as reporting what passed.

4. What public patents, papers, and media verify this track record?

Because client consulting operates under NDA, technical competence is documented through public filings, monographs, open survey datasets, and independent editorial coverage.

5. How can teams contact or verify Amir Shachar?

Direct personal, academic, and advisory messages go straight to Amir Shachar (never a call center). For formal live scoring evaluation or speaking bookings, Riskernel provides structured intake channels.

6. Frequently Asked Questions

Who is Amir Shachar and what is his industry background?
Amir Shachar is the founder and Chief Scientist of Riskernel. Previously he served as Chief AI Scientist at Skyhawk Security (co-leading its AI purple team and authoring 3 granted cloud-security patents) and Head of Fraud Data Science at NICE Actimize, where he received the 2019 CEO Excellence Award for patented commercial deployment of federated learning across Tier-1 banks.
What academic institutions did Amir Shachar attend and teach at?
Amir Shachar studied Mathematics, Computer Science, and Statistics at the Technion - Israel Institute of Technology and Tel Aviv University. Between 2008 and 2010, he served as a Teaching Assistant at the Technion across four courses: Calculus 1, Advanced Calculus, Linear Algebra 1, and Image Processing.
What research frameworks and patents has Amir Shachar authored?
Amir Shachar is the named inventor on 14 patents (9 granted US patents and 5 pending filings from 2014 to 2026) and author of Introduction to Algogens (arXiv:2403.01426, 560 pages, 2024), Semi-discrete Calculus (arXiv:1012.5751, 2010, and the 808-entry Detachment Library), and Inquiry Policy Learning (2026).
Where can I verify Amir Shachar's public credentials and publications?
Public records are indexed directly on /facts/, /patents/ (linked to Google Patents and Justia), /publications/ (linked to arXiv), /media/ (6 press features in Dark Reading, CSO, SC Media, TechTarget, startuphub.ai, and Yahoo Finance), and external profiles on LinkedIn, Google Scholar, GitHub, Academia.edu, and ResearchGate.
How do teams contact Amir Shachar for fraud AI advisory or speaking?
Direct personal, research, and advisory inquiries go to contact@amirshachar.com. Formal Riskernel advisory sessions, Launch Fraud Reviews, shadow testing, and conference speaking requests can also be initiated at riskernel.com/advisory.html and riskernel.com/speaking.html.