CV · LinkedIn · erenomer@gmail.com

Twenty years of building software, still writing code every day. I work at the intersection of regulated consumer finance and applied artificial intelligence: credit decisions, money movement, and the systems that have to be auditable when someone asks why.

Currently at Türk Finansman (Kredim), building the systems behind credit decisions.

SkorAI, credit scoring and decision support Modular scoring platform in Flask with a YAML rule engine and a React and TypeScript management interface. Business rules live in a central configuration store instead of the codebase: reject, manual-review and flag rules are versioned into an audit trail, so a policy change is a reviewable record rather than a deploy. After the engine went live I ran the comparative before-and-after risk and limit analysis across roughly one million customers.

AML sanctions screening In-house screening platform built as Rust and Go microservices over PostgreSQL and OpenSearch. Ingests official sanctions lists, matches customers with recall and rerank scoring, and routes hits into case workflows.

Local retrieval-augmented generation (RAG) service Vector-based knowledge retrieval that runs on premises, so company data never leaves the building, connected to the team's language model tools over the Model Context Protocol (MCP).

Format-preserving anonymization platform Written in Rust. Masks personal data while keeping referential integrity across systems, so anonymized copies stay joinable.

Complex event processing (CEP) engine · Monachus, 2021-2025 Multi-component engine for rule-based decisions on streaming data. The project received public innovation funding through KOSGEB.

OpenHub · open source Multi-agent orchestration: distributes work to AI agents, runs results through verification gates, and waits for human approval at the critical steps.

Python TypeScript Rust Go Java C# SQL / T-SQL FastAPI Flask Spring Boot Node.js .NET React Next.js React Native PostgreSQL SQL Server OpenSearch Redis Docker Kafka LLM applications RAG MCP

_{Most of the work above lives in private company repositories. What is public here is side work, tooling and forks I keep for reference.}