Company
We verify the numbers institutions bet on.
Scientific Analytics Alliance builds independent risk infrastructure: a deterministic engine that recomputes what your existing systems report – and proves every result with a replayable evidence trail. We sit beside Aladdin, Bloomberg, Murex and MSCI. We replace none of them. We check all of them.
What we believe
Evidence before assertion.
The site should tell the truth quickly: what is measured, what is pending, what is narrative, and what no one should claim before the evidence exists.
Every number – fair value, VaR, CVaR, scenarios, verdicts – is computed by a deterministic kernel. Same input, same build, same result: one identical hash across 1,000 fresh runs.
AI writes narrative prose only, and can be switched off entirely – reports still render from computed figures. No AI-generated valuations, risk metrics or verdicts. Ever.
We hold no trading book and grant no exclusive licences. Customers’ books are our validation target, never our competition. The methodology is published, not patented.
We publish what we cannot claim next to what we can. Misses ship as WATCH, not silence. Certifications we do not hold are listed as roadmap, not implied.
Founder
Oleksii Slieptsov, CFA
Founder · CEO · Principal Architect
Nineteen years in treasury and liquidity at PrivatBank, Ukraine’s largest systemically important bank – including the role of Head of Interbank Cash Operations and NBU Cash Transactions Department, PrivatBank JSC CB from 2014, and uninterrupted national cash operations through wartime. Nineteen years as a daily Bloomberg Terminal and Thomson Reuters / Refinitiv Eikon user. In 2024 he began building what he believes those workstations should have become: a risk platform where every number can be independently verified. The framework behind the kernel is published under his own name – 16 working papers on SSRN as of Aug 2026.
What we build
One kernel, one control discipline, seven product surfaces.
ARIN22 is the compute spine across the domain modules. ARIN interprets and governs the narrative layer. KOKON turns the resulting evidence into company-control action.
The decision council: 20 domain agents, one Governor, one sealed verdict per query.
Open ->Equity and cross-asset research, CFA-charterholder-authored methodology.
Open ->Institutional AA-to-D proprietary analytic scores for on-chain assets – research labels, not credit ratings; 0% AI in scores.
Open ->
Proof
The numbers are the product.
Full numerics live under NDA. Public claims stay compressed, dated and bounded.
Evidence lanes as of Aug 2026 · brain graph snapshot updated 25 Aug 2026 · artifact-bound. Full numerics under NDA.
Research
We published our own examination.
DRCET – the Deterministic Risk-Compute Equivalence Test – is an open protocol for proving that a fast risk engine returns the Monte Carlo answer. DRCET-1 is published on SSRN; Deterministic Risk-Compute Equivalence Testing for Path-Dependent Functionals and Factor Attribution and Fail-Closed Release Gates for Numerical Operator Claims: Certificates, Observed Order, and Spectral Domains in Risk Computation were distributed August 2026. The protocol is deliberately unpatented: anyone can test any vendor, including us. A standard is worth more than a secret.
Trust posture
Pre-revenue. Transparent about it.
No undisclosed customer logos. Many claims are verifiable without asking: github.com/SAA-Alliance · sha256sum -c. Qualified diligence teams receive the audit pack for the remaining NDA-bound surfaces. Our validation lineage publishes the failures alongside the passes – the battery record includes every intermediate FAIL. Delivery is phased, and binding production promotion is not allowed until each phase clears its closure contract – that rule is enforced in the system, not in a policy document. Formal certifications (SOC 2 and others) are on the post-revenue roadmap and are not claimed today.
Engagement
Commercial paths stay bounded.
Pricing is disclosed under signed engagement. Every model ships with an explicit scope boundary and claim matrix.
One portfolio, challenger report, divergence analysis and evidence pack.
Writing: scope boundary + claim matrixOne domain: banking, insurance, sovereign, real estate or asset management.
Writing: module boundary + evidence laneMulti-domain stack with API, monitoring and evidence workflows.
Writing: control matrix + promotion gatesPrivate deployment for governments, critical infrastructure and banks.
Writing: data boundary + deployment postureARIN22 as the compute layer inside existing risk systems.
Writing: API scope + refusal conditions