8000 ahmettugur (Ahmet Tügür) · GitHub
[go: up one dir, main page]

Skip to content
View ahmettugur's full-sized avatar

Block or report ahmettugur

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
ahmettugur/README.md

Typing SVG


🧭 About me

public sealed record Developer
{
    public string Name      => "Ahmet Tügür";
    public string Role      => "Senior .NET Developer";
    public string[] Focus   => ["Multi-Agent AI Systems", "Distributed Architecture", "Observability"];

    public string[] Patterns => [
        "Clean / Hexagonal / Onion Architecture", "DDD", "CQRS", "Event Sourcing",
        "ReAct + Self-Reflection", "Human-in-the-Loop", "RAG & Semantic Memory"
    ];

    public string[] Stack   => ["C#", ".NET 10", "Microsoft Agent Framework", "Semantic Kernel",
                                "Kafka", "RabbitMQ", "PostgreSQL", "Qdrant", "Docker", "Kubernetes"];

    public string Belief    => "An LLM is just another adapter — the architecture still matters.";
}
  • 🤖 Currently building production-grade agentic systems: specialized agents, orchestration, tool routing, HITL approval gates and self-improving feedback loops
  • 🏗️ Long-time practitioner of Clean / Hexagonal Architecture, DDD, CQRS and Event Sourcing — in both .NET and Spring Boot
  • 🔭 Observability-first: every agent step, LLM call and tool invocation is traced with OpenTelemetry
  • 🎓 Mentor to interns and junior developers — I write training material on architecture and agentic AI

🌟 Flagship: Agentic Customer Support Bot

A multi-agent customer support system built with Microsoft Agent Framework (MAF) on .NET 10, structured as Hexagonal Architecture (Domain · Application · Adapters).

flowchart LR
    U([👤 Customer<br/>Text · Voice]) -->|SSE stream| P[🧭 PlanningAgent]
    P --> PR[📦 ProductAgent]
    P --> OR[🚚 OrderAgent]
    P --> CO[📝 ComplaintAgent]
    PR & OR -.->|parallel, read-only| R
    CO -->|side effect| H{🙋 HITL<br/>approval}
    OR -->|place order| H
    H --> R[✍️ ResponseAgent<br/>self-critique]
    R --> U
    R --> T[(📈 OpenTelemetry<br/>traces + cost)]
    T --> L[🔁 Self-Improving Loop]
    L -->|admin-approved lessons| M[(🧠 Qdrant<br/>RAG + episodic memory)]
    M -.-> P
Loading

🧠 Reasoning & Control

  • Structured reasoning — preToolCheck / postToolReflection (ReAct + Self-Reflection)
  • Human-in-the-Loop gates for side-effecting actions
  • Parallel sub-task execution for compound queries (Task.WhenAll)
  • SLA Guardian: auto-reject stale approvals, auto-escalate priority

📚 Memory & Learning

  • Semantic memory + RAG on Qdrant, plus episodic memory
  • Per-customer personalization profile injected via a context pipeline
  • Self-improving loop: low-rated traces → LLM-generated lessons → knowledge base
  • Skills-based smart routing for human escalation

🎤 Realtime Voice (dual mode)

  • Bridge mode: voice in/out over the full 6-agent pipeline, HITL intact
  • Native mode: model-side function calling on read-only tools — ~3–5× faster

🔭 Operability

  • OTLP traces & metrics → Jaeger / Prometheus / Grafana
  • Per-model USD cost & token accounting
  • Trace Replay UI to step through any workflow run

→ Explore the repository


🚀 Featured projects

🤖 AI & Agents

Project Highlights Stack
agentic-customer-support-bot 6-agent MAF pipeline, HITL, RAG + episodic memory, self-improving loop, realtime voice, trace replay .NET 10 MAF Blazor WASM PostgreSQL Qdrant OpenTelemetry
enterprise-ai-assistant AI assistant for business intelligence: intent routing to Chat / Document / Report agents, natural-language → SQL reports, document RAG, scheduled jobs .NET 10 Semantic Kernel Angular SignalR Neo4j Qdrant DuckDB Hangfire

🏛️ Architecture & Distributed Systems

Project Highlights Stack
MinimalApiCleanArchitecture Minimal API + gRPC on Onion Architecture; centralized logging, distributed tracing, service discovery, k8s deployment, SonarCloud quality gates .NET 8 gRPC Serilog ELK RabbitMQ Jaeger Zipkin Consul
CQRS-EventSourcing Banking account domain with separate command/query APIs and a pluggable broker (Kafka or RabbitMQ) .NET 8 Kafka RabbitMQ MongoDB SQL Server
SpringBoot-CQRS-EventSourcing The same CQRS + ES ideas on the JVM — reusable cqrs-es core, command / query services Spring Boot Kafka MongoDB MySQL PostgreSQL
AspNetAspireMicroservice Cloud-native microservices orchestrated by .NET Aspire with built-in tracing, structured logs and dashboard .NET 9 Aspire EF Core PostgreSQL RabbitMQ OpenTelemetry

☁️ Cloud Native & DevOps

Project Highlights Stack
devops-shopping .NET microservices deployed to Azure Kubernetes Service with automated Azure Pipelines AKS Azure DevOps Docker k8s
ProductsStoreOnKubernetes ASP.NET MVC + SQL Server on Kubernetes with Helm charts, local & Azure manifests Kubernetes Helm SQL Server Azure
🧰 More: Go, full-stack samples & case studies
Project Description
toolkit Reusable Go module — JSON I/O, file upload/download, slugs, random strings (go get github.com/ahmettugur/toolkit)
TechCase Add-to-cart technical case with Redis, SQL Server, Docker and tests
dotnet-core-webapi ASP.NET Core Web API sample
react-redux-aspnet-core-webapi · react-redux-golang-webapi · react-redux-nodejs-webapi Same React + Redux frontend against three different backends — .NET, Go and Node.js

🛠️ Tech stack

Languages & Frameworks

Data, Messaging & Search

Cloud, DevOps & Observability

AI & Agents

🧩 Architecture & patterns I work with
Area Patterns & practices
Application architecture Clean · Hexagonal (Ports & Adapters) · Onion · Modular Monolith · Vertical Slices
Domain DDD (aggregates, domain events, bounded contexts)
Distributed systems CQRS · Event Sourcing · Outbox · Saga · Service Discovery · gRPC
Agentic AI Planner / specialist / responder topology · ReAct · Self-Reflection · HITL · RAG · Episodic memory · Tool routing
Observability OpenTelemetry traces & metrics · Jaeger / Zipkin · Serilog + ELK · LLM cost telemetry
Delivery Docker · Kubernetes · Helm · Azure DevOps · .NET Aspire · SonarCloud

📊 GitHub activity

GitHub stats Top languages

GitHub streak

Contribution snake


"Architecture is about the important stuff — whatever that is." — Ralph Johnson

Pinned Loading

  1. agentic-customer-support-bot agentic-customer-support-bot Public

    Agentic customer support system using .NET 10 + Microsoft Agent Framework (MAF). Features 6-agent orchestration (Planning + Specialists + Response), deterministic reasoning pipeline, HITL approval …

    C# 5

  2. enterprise-ai-assistant enterprise-ai-assistant Public

    Enterprise-grade multi-modal AI assistant powered by .NET 10, Angular 21, Semantic Kernel and SignalR. Features Hexagonal Architecture, Multi-Agent SQL pipeline, Hybrid RAG (Qdrant + HyDE), Neo4j S…

    C# 2

  3. AspNetAspireMicroservice AspNetAspireMicroservice Public

    Asp.Net Aspire Microservices .Net 9

    C# 3

  4. MinimalApiCleanArchitecture MinimalApiCleanArchitecture Public

    minimal-api-clean-architecture .NET9

    C# 22 7

  5. CQRS-EventSourcing CQRS-EventSourcing Public

    CQRS & Event Sourcing with Kafka, RabbitMQ, MongoDB and .NET 8

    C# 4

  6. react-redux-golang-webapi react-redux-golang-webapi Public

    JavaScript 28 2

0