Oleksandr S.

Project overview:

Architected and developed a sophisticated loan application processing platform with agentic AI capabilities for a fintech lending company. The system combines an N8N-inspired declarative workflow engine with LLM-powered agents equipped with 30+ MCP tools for autonomous task execution. Core modules include automated bank statement extraction and fraud detection, vector-powered semantic search for document analysis, and multi-tenant architecture with complete data isolation across 76 database tables.

Responsibilities:

• Defined end-to-end system architecture: designed the workflow engine model, agentic orchestration layer, financial processing pipeline, and vector search infrastructure as cohesive modules with clear contract boundaries.
• Architected an N8N-inspired workflow engine with 15+ node types, declarative JSON-based workflow definitions, and configurable execution states.
• Designed and implemented agentic AI orchestration layer with 30+ MCP (Model Context Protocol) tools covering browser automation, business data retrieval, financial processing, and web search/deep research capabilities.
• Designed financial processing module: bank statement PDF extraction, 2-pass LLM transaction classification with confidence scoring, automated fraud detection, and configurable knockout/manual review decision logic.
• Architected vector search infrastructure with RAG pipeline, defined embedding and chunking strategies for semantic document analysis and retrieval using Weaviate.
• Designed multi-tenant data architecture with JSONB-based flexible entity storage (76 tables), event- driven audit system, and webhook notification pipeline.
• Built comprehensive observability layer: LLM metrics tracking, workflow audit trails, structured logging, and dynamic configuration management.

Achievements:

The platform automated complex loan underwriting workflows, significantly reducing manual review time through configurable knockout and auto-approval thresholds. The agentic AI system autonomously handles multi-step research and verification tasks using 30+ MCP tools spanning browser automation, financial processing, and web search. A 2-pass LLM transaction classification pipeline with confidence scoring improved categorization accuracy. Multi-provider LLM architecture enables seamless switching between five providers without code changes.

Technology stack:

Project overview:

Led an AI-first transformation of the software development lifecycle for a charity and volunteering platform serving enterprise clients. The initiative encompassed driving AI code-generation adoption across the engineering team, building AI-powered internal tooling integrated into sprint workflows, implementing MCP servers for streamlined AI-assisted development, and constructing a RAG-powered knowledge base to surface institutional knowledge at scale. The project spanned full-cycle development including requirements, architecture, security, and policy integrations using an AI-first approach.

Responsibilities:

• Designed and executed a structured SDLC transformation strategy, driving measurable AI adoption metrics across the engineering team within a 5-month timeline.
• Built AI-powered internal tooling: a Spring Microservice Factory for automated service scaffolding and an AI-assisted test-case generator integrated directly into sprint workflows.
• Implemented MCP servers to streamline AI-assisted development; integrated LangChain4j, Koog, and SpringAI into production services for agentic capabilities.
• Built an AI-powered internal knowledge base with RAG architecture to surface institutional knowledge at scale, cutting onboarding time by 66%.
• Led full-cycle development (requirements, architecture, security, policy integrations) using AI-first approach with Cursor and Claude Code.
• Designed AWS-native architecture; executed ETL migration of legacy codebase and data to modern stack including Flyway-managed database migrations.
• Conducted hands-on knowledge-sharing sessions on AI agents and agentic solutions; collaborated with distributed teams across Canada, Spain, and Romania.

Achievements:

Drove AI code-generation adoption from 23% to 77% across the engineering team within 5 months. Reduced team onboarding time from 3 weeks to 1 week by building a structured project knowledge base with AI use-case guides. Decreased repetitive engineering work by 56% through custom AI-powered tooling. Raised team AI trust score from 2 to 4 out of 5 through hands-on knowledge-sharing sessions. Reduced time-to-delivery from approximately 1 month to 2 weeks through streamlined CI/CD and AI-assisted code review.

Technology stack:

Project overview:

Led cloud-native development and a team of 5 engineers for a smart consumer IoT platform processing real-time device telemetry from connected smart beds. The platform managed cloud infrastructure for hundreds of thousands of connected devices, encompassing microservice architecture, real-time observability, data lifecycle management, and compliance reporting across AWS-native services.

Responsibilities:

• Led a team of 5 engineers end-to-end, reducing time-to-production by 22% from high-level requirement to production release.
• Designed and implemented data archiving strategies and RDS layer redesign, achieving $12,000/month infrastructure cost reduction.
• Built a dynamic Datadog metrics library enabling real-time observability across all service layers with proactive alerting that prevented downtime and measurably improved customer satisfaction.
• Decoupled heavy synchronous compliance reporting into a dedicated async microservice, eliminating near-completion failures that previously required full restarts.
• Executed Java 8 to 11 to 17 migration across legacy microservices; modernised the codebase and removed deprecated dependencies with perfect SonarQube scores.
• Conducted security code reviews; mentored junior engineers. Active member of the technical interviewer community across 5 years. Collaborated with distributed teams across US, Canada, India, Pakistan, Slovenia, UK, and Ukraine.

Achievements:

Reduced monthly AWS infrastructure costs by $12,000 through RDS layer redesign and data archiving strategies. Decreased customer service requests by 38% via proactive alerting enabled by a custom dynamic Datadog metrics library. Eliminated an 80% failure rate on compliance report generation by decomposing a synchronous process into a dedicated async microservice. Achieved 100% test coverage on all new services and raised legacy code coverage from 33% to 70%. Reduced time-to-production by 22%.

Technology stack: