Data & AI strategy
Board-level data and AI strategy, roadmaps, building data organizations from the ground up — including the Fractional CDO / CDAIO format.
The practice
What we do
From a focused advisory engagement to owning the Data & AI agenda end to end.
Data platforms
Lakehouse architecture — cloud, on-premise or hybrid — consolidation of legacy estates, custom-built platforms where off-the-shelf won't do, data models and ontologies for regulated domains: from blueprint to production.
AI/ML & generative AI
Real-time ML pipelines, fraud and financial-crime detection, LLM and RAG in regulated environments, MLOps: monitoring, retraining, explainability.
Data governance
Governance operating models for regulated environments: policies and standards, data ownership and stewardship, quality rules enforced in the pipelines themselves, lineage and access control — audit-ready by design, proven in AML/CFT settings.
Analytics & BI
Data warehouses, marts, self-service reporting and graph analytics — the business gets its answers without queueing for IT.
Technology due diligence
Independent assessment of architecture, data assets, teams and technology risk — for investors, boards and incoming executives.
Expertise
Where we are at home
Domains
Financial monitoringAML / CFT / KYCFraud detectionGeospatial · NSDITax analyticsERP / CRM / HRM
AI & ML
Machine LearningGenerative AI · LLM · RAGAI agentsGraph analyticsNLPMLOps
Platforms
Data LakehouseDatabricksDremioDataikuApache KafkaApache SparkGraph databases
Standards
OGC / ISO 191xxSTACRegulatory requirementsModel auditabilityData governance & stewardshipData quality & lineage
Selected work
Proven at national scale
National scale
Financial monitoring with graph intelligence
Graph analytics makes hidden links between people, companies and accounts visible at once; ML cuts false positives; a domain LLM assists investigators — summarizing case context and answering questions in natural language. 100K+ transactions a day under continuous AML screening, with regulator-grade consistency of reporting.
Country-wide
A nation's geospatial data on one platform
National Spatial Data Infrastructure: a single catalog on open standards (OGC/ISO 191xx, STAC), self-service tools and APIs across ministries and agencies — less duplication, faster decisions.
Public finance
An analytical platform for a national tax authority
Scattered, unstructured data consolidated into a central warehouse with data marts: end users produce the reports they need themselves, in the format they need — no IT tickets.
Aviation
ERP and analytics for the region's largest airline
Fourteen legal entities unified into one structure with common financial processes; big-data analytics on flight and maintenance data for operational efficiency and data-driven planning.
Sports integrity
Athlete information system for AMADA, the national anti-doping agency
Registry, tracking and analytics in one system for Azerbaijan's National Anti-Doping Agency, with records anchored on a permissioned Hyperledger Iroha ledger — the country's first blockchain product developed by a local company. The same ledger-backed approach later reached agriculture: national seed certification on Algorand, the first blockchain deployment in the country's agrarian sector. On the record — independent press coverage.
Principles
How we work
A system, not a report
The deliverable is a platform in operation, not a presentation. Architecture, build and launch sit within one line of accountability.
Regulated by default
Auditability, model explainability, access control and regulatory requirements are designed in from day one — never bolted on.
Built to outlast us
MLOps, monitoring, documentation and handover to your team — the system keeps living and evolving after the engagement ends.
Leadership
Who stands behind it
Alexander Kuklin
Chief Data & AI Officer at Caspel, strategic advisor to the CEO of SetClapp, formerly CTO of a 110-engineer organization and at Oracle (EMEA President Club). Y Combinator alumnus. Twenty-five years of building data systems — from early warehouses to national AI platforms.
Delivery teams are assembled per engagement from a vetted senior network; architecture and accountability stay with the founder. Every Kuklin Pro engagement passes through his architectural review — from the first diagram to production acceptance.