API Endpoints
Shared Services
Distributed Systems Architect
I build software that stays reliable once it ships — clear boundaries, observable behavior, and systems made to last.
2+ years shipping production backend systems — across enterprise platforms, health infrastructure, and e-commerce serving millions of users.
I specialize in distributed architectures — designing the boundaries, messaging patterns, and data consistency strategies that make microservices predictable at scale.
Currently exploring new opportunities — open to senior backend and distributed systems roles where scale, reliability, and AI infrastructure are the hard problems.
Not demos. Not tutorials. Systems that process thousands of transactions daily, serve millions of users, and maintain uptime SLAs under real pressure.
API Endpoints
Shared Services
Designed Product tracking software system all the way through
Platform Type
Architecture
Vehicle tracking fleet software system for the owners and employees
Web Interface
Error Rate
Platform
Stack
E-commerce platform for a company that manufactures printing equipment
API Counts
Market
Multi Turkish Gaming platform
API Endpoints
Uptime
Engineering principles are not decisions made on the spur of the moment. They involve designing systems that allow for development from the very beginning.
A design that is as understandable as possible, with details serving as elements that add variety to this simplicity.
Technical debt is a constant in architectures that scale on demand. The infrastructural backbone established during the initial design phase — how data is partitioned, how state management is decoupled from services, and asynchronous data flow patterns — directly determines whether the system will withstand peak loads or fail.
Errors are inherently part of the system; those that indicate persistence, however, are a major problem.
Every software system that is designed will inevitably encounter issues. The real question is whether these issues are simple, fixable problems, or part of an ongoing chain of failures. Concepts such as bulkheads, circuit breakers, and graceful degradation help us determine whether a situation is a full-blown disaster or something that could have been prevented.
Traces, metrics, and logs are essential for any piece of software.
Observability is not a plugin to be integrated after going live; it is a design constraint that must be followed from day one. Structured logs, end-to-end distributed tracing, and RED metrics transform an opaque black box into a transparent and manageable system.
The best architecture is the one you can explain and design.
The quality of a system is measured by how explainable and manageable it is. CQRS, microservice architecture, and event-driven structures are not a destination — they are methods. Every abstraction without a sound reason adds weight. A plain monolithic design on solid ground is stronger than a service mesh that has lost control and is full of blind spots.
Not using AI as a feature — integrating intelligence as infrastructure. Embedding pipelines, agent orchestration, and local inference as production concerns.
Retrieval-Augmented Generation systems connecting LLMs to proprietary knowledge bases. Embedding pipelines, vector store integrations, and context-window management for domain-specific applications.
Multi-agent systems where specialized agents handle distinct tasks coordinated by an orchestrator. Built for customer support automation and internal tooling with function calling and tool use.
On-premise LLM deployment using Ollama for air-gapped environments. Evaluated Mistral, LLaMA, and Phi families for domain-specific tasks with GGUF quantization strategies.
Designing .NET backend services that orchestrate AI workflows as first-class infrastructure. Streaming responses, async queuing for inference, and cost-aware routing between providers.
Retrieval-augmented generation: from a user query to a grounded response.
Not just built to work — built to keep working when things go wrong. Event-driven order pipeline: how messages flow, retries happen, and consistency holds without tight coupling.
If you're building systems that need to survive production — distributed, resilient, and actually shipped — let's talk.