Yavuz Çelik

Full Stack .NET Developer

A Full Stack developer working mainly on the backend with C# and .NET. I build web, mobile and desktop applications, and integrations with ERP systems.

NOW

Research Engineer · Sakarya Üniversitesi

Full Stack .NET Developer · Esbi Bilişim ve Telekomünikasyon

FOCUS
Backend architecture · ERP integrations · Document AI and OCR
  1. CLIENTSweb · mobile · desktop
  2. APIsignalr · web api
  3. SERVICESapplication core · ocr pipeline · background workers
  4. DATAsql server
  5. EXTERNAL SYSTEMSlogo erp · payment providers · bay.t erp · marketplaces
FIG. 1The layers I work across: clients, API, services, data, and the external systems they connect to.

01 / ABOUT

What I do

A short introduction and the areas I work in.

Portrait of Yavuz Çelik

I'm a Full Stack developer focused on backend development with C# and .NET. Besides building web, mobile and desktop applications, I work on integrations with ERP systems, payment providers and external services. I'm currently pursuing an M.Sc. in Data Science and Artificial Intelligence at Sakarya University; my research is on FormReader, a system that extracts structured data from handwritten Turkish production forms using OCR, computer vision and vision-language models. I'm interested in backend architecture, document AI, OCR and applied machine learning.

TürkiyeEsbi Bilişim

Years of Experience
3+
Production Projects
20+
Users Served
15K+

02 / CAPABILITIES

What I work with

The areas I work in, and the tools I use in each.

01Backend Engineering

Business applications and APIs on ASP.NET Core, including multi-tenant systems.

  • C#
  • ASP.NET Core
  • .NET 8
  • ASP.NET Web API
  • SignalR
  • Clean Architecture
02Systems & Integrations

Connecting applications to ERP systems, payment providers and marketplaces over REST, SOAP/WCF and SQL.

  • Logo ERP
  • Bay.t ERP
  • REST API
  • SOAP / WCF
  • RabbitMQ
  • Marketplace APIs
03Data & Persistence

Relational schemas and query work on SQL Server, with EF Core and Dapper side by side.

  • SQL Server
  • PostgreSQL
  • EF Core
  • Dapper
  • Redis
04Frontend

Web interfaces with Razor and React.

  • TypeScript
  • React
  • Next.js
  • Tailwind CSS
  • Razor
05Mobile

A field-sales app built with React Native and Expo.

  • React Native
  • Expo
06Desktop

WinForms tools and Windows services that run at customer sites.

  • WinForms
  • DevExpress
  • Windows Service
07Document AI & Computer Vision

OCR and computer vision pipelines for handwritten forms; the subject of my ongoing M.Sc. research.

  • Python
  • OpenCV
  • PaddleOCR
  • Computer Vision
  • Vision-Language Models
08DevOps & Delivery

Build and deployment pipelines with GitHub Actions; hosting on IIS.

  • Docker
  • GitHub Actions
  • IIS
  • Git
09Security

Authentication, password hashing, encryption and rate limiting inside the applications above.

  • JWT
  • PBKDF2 / BCrypt
  • AES
  • Rate Limiting

03 / PROJECTS

Selected work

Some of the things I have worked on.

01 / ACADEMIC

Research · accepted at ICEECIT 2026 · ongoing

FormReader — Extracting Data from Handwritten Production Forms

A hybrid OCR system that turns handwritten Turkish production forms into structured, reviewable data. It is the subject of my M.Sc. research at Sakarya University.

  1. Scanned formhandwritten · Turkish
  2. Geometric normalizationtemplate-first
  3. Cell extractioncanonical cells
  4. RecognitionSVTR + CTC · VLM for free text
  5. Quality layerauto-accept · human review
  6. Structured dataExcel · JSON
FIG. 2FormReader pipeline: from a scanned handwritten form to structured data. The quality layer sends each field to automatic acceptance or human review.

PROBLEM

In manufacturing plants, downtime durations, downtime causes and scrap counts are still written onto paper forms by hand and retyped into a spreadsheet later. Because these values feed downtime and OEE calculations, a wrong value written silently to a database costs more than a rejected one.

SOLUTION

I built a hybrid pipeline that routes each field to a recognition path suited to its type and attaches a quality decision to every output. The form is first aligned to its template and fields are read from their own cells; a quality layer then sends each value either to automatic acceptance or to a human review queue. The system does not promise guaranteed accuracy; it measures its uncertainty.

  • Python
  • PaddleOCR
  • SVTR + CTC
  • OpenCV
  • Vision-Language Models
  • QLoRA
  • React
  • .NET
  • PostgreSQL
Read the case study

02 / WEB

.NET 8 · multi-tenant · ERP-integrated

Multi-Tenant B2B/B2C E-Commerce Platform

An ERP-integrated B2B/B2C e-commerce platform for companies with dealer networks. Derived from a single core product and adapted for four companies in different industries.

  1. MarketplacesTrendyol · Hepsiburada · N11 · Shopify
    Payment providersIyzico · PayTR · Param · Paynet · QNB · Tami · e-Tahsilat
    Client applicationsB2B · B2C
  2. Application core.NET 8 · ASP.NET Core MVC · multi-tenant
  3. ERP abstraction layerLogo and Bay.t in the same codebase
  4. Logo ERPUnityObjects (COM) · WCF facade
    Bay.t ERP
FIG. 3Marketplaces, payment providers and client applications meet in one core; ERP differences stay behind the abstraction layer.

PROBLEM

Companies with dealer networks across different industries (motorcycle parts, carpets, tractors, textile) took orders by phone and Excel, then entered them into their ERP by hand. Each client had its own price lists, stock rules and payment flows.

SOLUTION

I built a multi-tenant B2B/B2C platform on .NET 8, derived from a single core product and customized per client. I designed an abstraction that supports both Logo and Bay.t ERPs in the same codebase, and automated the order-to-invoice cycle.

  • .NET 8
  • ASP.NET Core MVC
  • EF Core
  • Dapper
  • SQL Server
  • SignalR
  • WCF
Read the case study

03 / WEB

ASP.NET Core 8 · multi-tenant · relay architecture

ESBI-Mutabakat — Account Reconciliation SaaS

A multi-tenant SaaS that digitalizes account reconciliation for companies running LOGO/Bay.t ERPs.

  1. Counterpartytokenized approval page
  2. Cloud SaaSASP.NET Core 8 · multi-tenant
  3. CUSTOMER NETWORK
    1. Relay agentESBI-Connect · allow-list
    2. ERP databaseLOGO · Bay.t
FIG. 4ERP data is reached through an allow-listed relay inside the customer's network; the database password never leaves the company.

PROBLEM

Account reconciliation was a slow, legally required process where accounting teams emailed hundreds of counterparties by hand every month and tracked replies in Excel. Handing out the ERP database password was also a security risk.

SOLUTION

I built a multi-tenant SaaS with ASP.NET Core 8 and Clean Architecture. Companies email balance/statement reconciliations to their counterparties, who approve or dispute them through a tokenized page. ERP data is accessed through an allow-listed relay agent, so the database password never leaves the company.

  • ASP.NET Core 8
  • Clean Architecture
  • EF Core
  • DevExtreme
  • QuestPDF
  • SignalR
  • Tailwind CSS
Read the case study

04 / MOBILE

React Native + Expo · GPS and push · LOGO-integrated

ESBI-Visitor — Field Sales Automation

A mobile app giving field sales teams GPS-stamped visit tracking, push notifications and LOGO account statements.

  1. Mobile appReact Native · Expo · GPS
  2. Web APIASP.NET Core 8
    Push queuebackground service · Expo Push
  3. SQL ServerEF Core
    LOGO ERPaccount statements
FIG. 5Visits are GPS-stamped on the device; the API serves LOGO account statements and queues push notifications.

PROBLEM

Field sales reps noted visits on paper and called the office to learn customer balances. Managers had no visibility into who was where in the field.

SOLUTION

I built a mobile app with React Native 0.81 + Expo 54 (TypeScript) and a JWT-protected Web API on ASP.NET Core 8. Visits are GPS-stamped, manager assignments are delivered via server-side Expo Push, and the customer's LOGO account statement can be viewed in the field.

  • React Native
  • Expo
  • TypeScript
  • ASP.NET Core 8
  • JWT
  • EF Core
  • SQL Server
Read the case study

All projectsMore on GitHub

04 / EXPERIENCE

Where I have worked

  1. Jan 2026 — Present · 10 mos

    Research Engineer

    Sakarya ÜniversitesiTürkiye

    I develop FormReader, a system that extracts structured data from handwritten Turkish production forms. I work on template-first geometric normalization, PaddleOCR/SVTR-based recognition models, vision-language models for free-text fields and a human-in-the-loop quality layer, and evaluate them on writers unseen during training. A paper on the work has been accepted at ICEECIT 2026. The research is supported by Sakarya University BAP.

  2. Sep 2023 — Present · 3 yrs 2 mos

    Full Stack .NET Developer

    Esbi Bilişim ve TelekomünikasyonTürkiye

    I develop and maintain B2B/B2C business applications using C#, .NET Core and related technologies. I build REST API integrations between ERP systems, marketplaces, payment providers and internal applications, and also work on a field-sales mobile app with React Native/Expo and on CRM, reporting and invoice management applications with Windows Forms and DevExpress.

  3. Jun 2022 — Jul 2022 · 2 mos

    Software Intern

    Kurumsal Yazılım A.Ş. (HarmonyERP)Türkiye

    Developed C#-based desktop tools for processing financial documents such as receipts and invoices. I worked with OCR and image-processing techniques to extract and validate information from document images, and integrated the extracted data with existing database workflows.

05 / CONTACT

Get in touch

Feel free to write about questions, collaboration or job opportunities.