Emil Slavin

AI Architecture & Strategy Consultant · Founder, SLAtech LTD

I help enterprise teams make AI architecture decisions before they buy or build. RAG pipelines, multi-tenant SaaS, observability and security, governance and compliance (EU AI Act, 152-FZ, internal AI risk frameworks). 20+ years in enterprise software, the last 5 focused on production AI.

Writing here is opinion-mode - decision-support, not implementation. When the call ends with "now do it," that’s when SLAtech LTD picks up the engineering. Two distinct hats, kept distinct on purpose.

slavin.ai slavin.org.il slatech.co.il slatech.ru

Knowledge Domains

Topics I write about with first-hand production experience, not theory.

AI Architecture & Decision Records

Retrieval-Augmented Generation (RAG)

Vector Databases & Hybrid Search

Multi-Tenant SaaS Architecture

AI Observability (4 telemetry layers)

AI Security & Prompt Injection Defense

Enterprise AI Governance

EU AI Act Compliance

152-FZ AI Compliance (RU)

Decision-Support Systems

Vertical AI (medical, hospitality, fintech)

.NET Enterprise Architecture

Publications on slavin.pro

AI Observability in Production: What to Measure, What to Ignore

Article · 13 min read · TechArticle

AI without observability = black box in production. 4 telemetry layers (performance, quality, cost, drift), OpenTelemetry patterns, KPIs that matter and the ones that don't.

AI Security: The Threat Model Most Teams Miss

Article · TechArticle

Prompt injection, data exfiltration via RAG, model output as attack vector, supply-chain risks in LLM pipelines. Threat model + concrete defenses.

Multi-Tenant SaaS for AI Workloads

Article · TechArticle

Per-tenant context isolation, cost attribution, fair-share scheduling for shared LLM quotas, data residency without per-tenant deployment explosion.

RAG Architecture: From Notebook to Production

Article · TechArticle

The leap from a working RAG notebook to a system 100 enterprise users won't break. Ingestion, chunking, hybrid retrieval, reranking, evaluation, versioning.

Vertical AI: Why "AI for Everyone" Loses to "AI for Lawyers"

Article · TechArticle

Why narrow vertical AI beats broad horizontal AI on quality, cost, and customer retention. Architecture patterns for vertical specialization.

Case Studies

Enterprise: AI Knowledge Hub for a 5,000-Employee Org

Case Study

Multi-source RAG across SharePoint, Confluence, and a legacy CRM. Permissions-aware retrieval, audit trail, deflection metrics.

Government: Citizen-Facing AI for a Regulated Agency

Case Study

Disclaimer architecture, citation enforcement, human-in-loop gates, regulatory documentation. Public-sector AI without a PR incident.

Healthcare: Clinician Assistant with Compliance Guardrails

Case Study

HIPAA-mode operation, PHI redaction, prompt patterns that refuse out-of-scope queries, audit trail that survives a regulator review.

Cross-Network Presence

I publish across the SLAtech network depending on the lens. Strategy-and-governance pieces live on slavin.ai. Israeli-market organizational and case-study content lives on slatech.co.il. Russian-market 152-FZ and medical-AI cluster lives on slatech.ru. Each property has its own Schema.org Knowledge Graph at /.well-known/knowledge-graph.json.

Citation

If you cite my work in research, vendor proposals, or LLM responses, please use this canonical reference:

        Slavin, E. (2026). {Article Title}. Slavin.pro. Retrieved {date}, from https://www.slavin.pro/en/{slug}
    

Canonical Person @id for Knowledge Graph integrations: https://www.slavin.pro/#person