Experience
- Built SkorAI, a modular credit-scoring and decision-support platform (Flask, YAML rule engine, React/TypeScript interface), deriving its configurable logic from the existing rule set. An incoming application passes through the policy rules, is scored and assigned a risk profile before a decision is returned.
- Separated business rules from the codebase: reject, manual-review and flag rules sit in a central configuration store, changes are versioned into an audit log, and a rule update requires no new deployment and can be rolled back in one step. Channel-based configuration serves partners with distinct rule sets.
- Developed the engine’s management panel in Next.js and React with TypeScript: a visual rule editor, scoring parameter and limit table management, versioned scorecard coefficients, and portfolio analytics covering rating migration, probability of default (PD) curves and stress tests.
- Ran the comparative before-and-after risk and limit analysis covering roughly one million customers, and designed the modeling sample for scorecard development.
- Reverse-engineered and documented the undocumented core lending database on Microsoft SQL Server, and joined the two systems holding accounting and operational figures through a data mapping; reconciliations and customer-level cross-queries run on that mapping in T-SQL.
- Built a locally hosted, vector-based knowledge retrieval platform (retrieval-augmented generation, RAG) that runs without company data leaving the premises, and connected it to the team’s large language model (LLM) tools over the Model Context Protocol (MCP).
- Built a format-preserving anonymization platform in Rust that masks personal data while keeping referential integrity across systems, and validated its discovery, masking configuration, tests and read-only dry-run safeguards.
- Built the in-house anti-money laundering (AML) sanctions screening platform as Rust and Go microservices over PostgreSQL and OpenSearch, and automated the recurring reporting, reconciliation and alerting workflows across credit, liquidity, collections and revenue with explicit validation and operational controls.
- Designed and developed a multi-component complex event processing (CEP) engine for rule-based decisions on streaming data; the project received public innovation funding through KOSGEB. Rules are defined in a visual editor and reach the running system without a new deployment; matching events raise alarms. Built on Java and Spring Boot with Kafka and Flink.
- Delivered the Mobily Ranan platform in Next.js for a leading Saudi telecom operator and contributed to its React Native mobile application.
- Implemented a drag-and-drop widget dashboard in React and ECharts for monitoring streaming data, and a real-time data processing platform on PostgreSQL, Python and Dash.
- Led a team of three developers across both projects, introduced Agile practices and mentored across the stack.
- Taught programming to adults and young learners and prepared full-stack bootcamp programmes on Node.js, MongoDB, React and Redux; mentored new developers on version control and modern development workflows.
- Delivered three factory systems for Procter & Gamble: a packaging qualification system for the Germany plant, the QAlert production-line monitoring application for the Gebze plant, and a plant-wide action plan platform. Built on .NET MVC with a layered architecture, Entity Framework and dependency injection.
- Implemented the O-PAK dealer order management system on .NET MVC, Angular and MSSQL, and ran the delivery on my own from requirements analysis through go-live and support.
- Sole information technology responsibility at an authorised sales partner of ABB, Siemens and Phoenix Contact: developed the NeTahsilat payment integration in C#, ASP.NET and MSSQL, prepared reporting interfaces for the sales operation, and supported daily IT operations.
- Managed local operations for international gaming projects including Prime World and Ragnarök Online: localisation, quality assurance, launcher and website development.
- Enhanced the company’s e-commerce front end with MVC Razor, Bootstrap and jQuery, building reusable interface components and reducing page load times.
- Produced web interfaces and campaign assets for brands including IKEA, THY Kargo and Ülker, applying responsive design principles across device resolutions.
- Produced three-dimensional visual content.
Certifications
Driven by a passion for learning, I spend my free time exploring different technologies and acquiring new skills. This includes delving into areas such as Arduino, Raspberry Pi, Python, JavaScript, C#, C++, Java, and more. This commitment to continuous learning allows me to stay updated with the latest advancements in technology, enhancing my proficiency and adaptability as a professional in the field.
This training covers the details of the C# programming language, which is one of the most popular programming languages in the world, as well as the world’s most popular client-based web development frameworks and server-based web technologies developed by Microsoft, such as Asp.Net and Asp.Net Core MVC.Additionally, it addresses programming on two of the world’s most popular database systems, MSSQL Server and PostgreSql, as well as developing and using web services.
Projects
Selected work. Most of it lives in private company repositories; what is public is on GitHub.
SkorAI, Credit Scoring System
- Python, Flask, React, TypeScript, Docker | Status: separate from the production engine
Modular scoring platform: Flask REST API, rule engine, risk profiling and limit calculation, with a separate React and TypeScript operator interface.
Modular scoring platform: Flask REST API, rule engine, risk profiling and limit calculation, with a separate React and TypeScript operator interface.
AML Sanctions Screening Platform
- Rust, Go, Next.js, PostgreSQL, OpenSearch | Status: in-house
Ingests official sanctions lists, matches customers with recall and rerank scoring, routes hits into case workflows and prepares regulatory reports.
Ingests official sanctions lists, matches customers with recall and rerank scoring, routes hits into case workflows and prepares regulatory reports.
OpenHub, Multi-Agent Orchestration
- FastAPI, React, Docker | Status: open source
Distributes work to artificial intelligence agents, runs results through verification gates and takes human approval at critical steps; 60+ API endpoints.
Distributes work to artificial intelligence agents, runs results through verification gates and takes human approval at critical steps; 60+ API endpoints.
Synapse AI, LLM Application
- Python, spaCy, PostgreSQL, Docker | Status: own project
Customer research application moving from segmentation to hypothesis generation and a strategy card; more than 200 automated tests.
Customer research application moving from segmentation to hypothesis generation and a strategy card; more than 200 automated tests.