Oleksandr
Semekha

AI Engineer / Systems Integrator

Implementation, development and integration of enterprise systems. AI engineering and deployment.

RAG systems, semantic search and document processing. ERP development, API integrations, data engineering and workflow automation.

Семеха Олександр Миколайович
9 yearsspecialising in ERP development
2 yearsbuilding AI solutions
30+ integrationsof systems and APIs
Profile
Role
AI Engineer / Systems Integrator — LLM systems, enterprise data, ERP
Education
University of Toronto (SCS) — DevOps CertificateKyiv Polytechnic Institute — Telecommunication Systems and Networks
Business
Business process management, management accounting and corporate finance — P&L, Cash Flow, multi-entity consolidation
Agile
Certified Scrum Master®, Certified Scrum Product Owner®, Project Manager
Technical
Microsoft AI & ML Engineering Professional Certificate · Google AI Professional Certificate · ITIL: Incident and Problem Management
Languages
Ukrainian, EnglishSpanish, German (basic level)
Specialisation

Programming and solution development

Custom development on top of enterprise systems and business processes: data models, background jobs and batch processing, access control and roles, reporting layers, document workflows and system-to-system integrations. Database and query optimisation — locating bottlenecks, partitioning and indexing strategies, speeding up processing on tables with tens of millions of rows.

ERPAccounting SystemsWMSCRMHRMSupply ChainTrade Management

Artificial intelligence

Building AI solutions on top of enterprise systems: from process analysis and architecture selection through to data integration, deployment and quality control. Typical scenarios — intelligent search across internal documents, automated processing of source records, agents for repetitive operations.

RAGsemantic searchpgvectorembeddingsdocument intelligenceagentic workflowsself-hosted LLM

Integrations and data exchange

Synchronisation, conversion and migration of data, integrations between the client's systems and external services.

APIe-commerce integrationsEDIbanking client systemsbatch processingJSONXML

IT project management (Agile/Scrum)

Requirements gathering and analysis, task decomposition. Strict deadline discipline, transparent reporting, Agile/Scrum delivery. Effort estimation, quality control and documentation of decisions.

analysiseffort estimationplanningwork acceptancedocumentation
Domain experience
Manufacturing & industrialCable and wire · Chemicals · Switchgear
Retail & consumerE-commerce · Apparel and homeware · Car parts · Office supplies · HoReCa
Logistics & customsCustoms brokerage · Customs clearance · Warehousing · National ticketing operator
Finance & reportingConsolidated reporting · Management accounting · Multi-entity groups
Regulated & asset-heavyHealthcare · Construction · Real estate · Property management
Experience
Track record
AI Engineer / Systems Integratormanagement company, group of 100+ legal entities2024 — present
ERP Developer / Integratorlogistics & trade2019 — 2024
ERP Developermanufacturing & customs2015 — 2019
Selected projects

IT & AI Governance Platform

multi-tenant SaaS built on a governance framework of 152 controls across 13 domains · solo, end-to-end · pre-revenue

Turns a maturity assessment — quick, full or AI-assisted — into a risk profile, a remediation plan and a complete, company-branded IT/AI documentation package: policies, orders, registers, NDA, DRP. Replaces 2–6 weeks of consultant work with minutes of generation (est.).

  • AI diagnostics — analyses the company profile, produces a risk assessment against the framework and turns it into an actionable remediation plan with a kanban board
  • RAG assistant — hybrid search (vector + full-text) over the 152-control corpus, answers strictly from the corpus with exact section citations; pgvector, Voyage embeddings, Gemini
  • Retrieval quality — 50-question eval set (25 in-scope, 25 adversarial out-of-scope): recall@5 100%, citation accuracy 100%, 21/25 correct refusals on adversarial prompts
  • Document engine — one single-source policy model generates 76 audit-ready PDF/DOCX artefacts with company details injected, each passing automated page-level verification
  • Audit & oversight — audit reporting per tenant, with a consolidated view for management companies across their portfolio
  • Engineering — multi-tenant PostgreSQL with RLS and group-level access, Next.js / FastAPI, GitHub Actions CI with 560+ automated tests
  • LLM operations — per-route observability: token usage and cost tracking, latency per call, model routing across embedding and generation models, full request log with admin dashboard
Loopdiagnostics → risk assessment → remediation plan → document generation → AI assistant → audit reporting

National ticketing operator — payment processing & document automation

designed and built end-to-end · production · multi-entity group
  • Payment ingestion — consolidates nationwide payment flows from banking systems and providers into a single accounting base; 10M+ row transaction tables, consolidated ledger across the group’s legal entities
  • API integration — two-way exchange with the operator’s ticketing API for order, payment and settlement data
  • Document automation — batch generation of orders, payments, delivery notes and tax documents, replacing manual per-transaction processing

Consolidated accounting & analytics for a management company

lead AI engineer · designed and built end-to-end · production · 100+ legal entities
  • Ingestion API & data store — custom API and MSSQL-based data store aggregating data from source systems across 100+ legal entities, including holding groups; Microsoft 365 mail intake as a secondary channel
  • Data quality & validation — cross-entity consistency checks over consolidated data before reporting
  • Analytics & reporting layer — consolidated reporting across investment projects and procurement, with a query layer built per holding group’s accounting model
  • AI assistant prototypes — bringing analytics and reporting into a single environment with an LLM assistant over consolidated data
  • Integrations — API integrations across multiple internal and external systems feeding the consolidated layer
Beyond engineering — led AI initiative selection: evaluating use cases, scoping projects and assessing vendors for delivery.

Document ingestion & OCR platform for customs clearance

designed and built end-to-end · production · 15+ enterprise clients, including US and EU-based
  • Ingestion & OCR — normalises invoices from PDF, XLS, XML, SAP IDoc and other formats into a canonical structure; OCR for scans, ~95% recognition accuracy in production
  • Validation engine — cross-checks weight, price, quantity, origin and description against reference data, catching declaration errors before submission
  • Auto-coding — HS-code assignment and translation over a historical declarations database, replacing manual classification
  • Document validity tracking — automated monitoring of certificate and document expiry dates across shipments
  • Integrations & reporting — two-way exchange with licensed customs declaration software; analytics over the full declarations archive
Part of a four-module logistics suite: tracking, customs, WMS, order management.
Tech stackAI & LLM Engineering: Python · FastAPI · LLM / RAG (Gemini, Azure OpenAI) · pgvector · embeddings · hybrid search · LLM evaluation · token cost & latency tracking · n8n · Microsoft FabricData & Enterprise Integrations: PostgreSQL (RLS, partitioning) · MSSQL (50M+ rows) · REST API · JSON / XML / EDI · enterprise ERP platforms · ERP integrationsProduct & Infrastructure: Next.js · TypeScript · Drizzle ORM · TanStack · Tailwind CSS · Docker · GitHub Actions