About and background

Data science delivery supported by a reliability-first engineering background.

Data Scientist and AI Engineer building evaluated machine-learning applications for fraud risk, financial forecasting, predictive maintenance, and computer vision, alongside LLM and RAG applications. I combine Python and model evaluation with SQL, tested data pipelines, APIs, and deployed portfolio demos. Results are published with their evaluation scope and limitations. My prior industrial automation work informs operational design; my applied data science training and projects are documented separately.

Verified experience

Verified project-delivery background.

Prior industrial automation roles remain documented as supporting domain context. Company projects are kept separate from personal data and software portfolio work.

May 2024 - July 2025

Project Engineer

Artı Endüstriyel Otomasyon Sistemleri

Istanbul, Turkey

Delivered software validation, SCADA/HMI revision, and industrial control-system work for refinery modernization and batch-production operations.

SOCAR - Heydar Aliyev Oil Refinery Modernization & Reconstruction

May 2024 - July 2025

Contributed to software validation and optimization for a large-scale refinery modernization project.

Executed Factory Acceptance Tests for Siemens PCS7 systems.
Improved SCADA HMI interfaces to strengthen operator situational awareness.
Verified data integrity across field devices and control systems.
Supported safe and reliable system deployment in a critical industrial environment.

Industrial AutomationSiemens PCS7SCADA/HMISoftware FAT

Marsa - Margarine Line Batch Control System HMI Revision

November 2024 - April 2025

Led the redesign of SCADA HMI interfaces for a batch-production system to improve usability and operational efficiency.

Redesigned legacy HMI screens for faster anomaly detection and a clearer operator experience.
Simplified batch-process visualization for more precise control.
Incorporated operator feedback to optimize production workflows.

Industrial AutomationSiemens PCS7SCADA/HMIBatch Control

Transferable context

Reliability-first validation, operator-centered interface design, and traceable delivery now inform how analytics and model outputs are reviewed before operational use.

January 2023 - April 2024

Automation Software Engineer

Dehatech Electric & Automation

Istanbul, Turkey

Designed and commissioned DCS/SCADA, virtualized IT/OT infrastructure, and industrial communication systems for international mining and zinc-processing facilities.

Iraq Waelz Facility: Zinc Recovery Automation & Infrastructure Design

January 2023 - April 2024

Designed and delivered the automation and IT/OT infrastructure for a greenfield zinc-recovery facility.

Built a virtualized server architecture using VMware ESXi for high availability.
Developed DCS control logic and operator interfaces using ABB 800xA.
Performed end-to-end system validation, including I/O testing and signal verification.
Delivered a ready-for-commissioning industrial control system.

Industrial AutomationABB 800xAVMware ESXiI/O Validation

Teksan Mining: Gas Analyzer Monitoring & SCADA Integration

December 2023 - January 2024

Designed and commissioned a SCADA-based gas-monitoring system for mining operations.

Developed ABB-based HMI interfaces for real-time monitoring of safety-critical data.
Integrated Modbus TCP/IP communication between analyzers, PLCs, and SCADA.
Configured server infrastructure for continuous data logging and monitoring.

Industrial AutomationSCADA/HMIABBModbus TCP/IP

Bulgaria Waelz Facility: Zinc Processing DCS Revamp & Commissioning

August 2023

Executed automation-system revamp and commissioning for a zinc-processing facility.

Refactored Siemens PCS7 and WinCC control logic for updated process requirements.
Conducted loop checks, I/O validation, and system testing.
Led commissioning activities including PID tuning and system optimization.

Industrial AutomationSiemens PCS7WinCCCommissioning

Transferable context

Architecture, system-boundary, data-flow, and validation experience strengthens the design of governed analytics workflows and operational decision-support systems.

February 2022 - January 2023

Project Engineer

KPI Automation

Istanbul, Turkey

Supported control-system modernization, validation, field integration, and commissioning for high-risk chemical-production operations.

Toros Tarım: Ammonia & DAP Production Lines DCS Revamp & Commissioning

September 2022

Led the modernization of control systems for Toros Tarım ammonia and DAP production lines.

Designed virtualized infrastructure using VMware ESXi for PCS7 and WinCC systems.
Executed full system validation focused on safety and interlock logic.
Managed field integration and commissioning processes.
Supported a seamless system transition with zero-downtime startup.

Industrial AutomationSiemens PCS7WinCCVMware ESXi

Transferable context

Safety-critical validation, commissioning discipline, and operational continuity provide the industrial context behind risk-ranked maintenance and decision-support work.

Skills and certifications

Modeling first, with the measurement discipline that makes it reviewable.

Machine learning and modeling

Pythonscikit-learnLightGBMXGBoostHistGradientBoostingRandom ForestSHAPFeature engineeringImbalanced classificationComputer visionOpenCVscikit-image

LLMs, AI engineering and deep learning

Gemini APIGenkitRAGLangChainMultimodal promptingTransformersTensorFlowKerasRNNs

Validation, calibration and measurement

Time-based holdoutWalk-forward with embargoGroupKFoldROC-AUCPR-AUCCalibrationThreshold simulation

Statistics and analysis

PandasNumPySciPyHypothesis testingT-testANOVAChi-squareCohort analysisRFM

Data engineering and automation

SQLBigQuerydbtDuckDBPostgreSQLMongoDBETL/ELTFivetrann8nZapierData quality testing

BI, MLOps and delivery

Power BIDAXLooker StudioGA4Google Tag ManagerDockerMLflowFastAPIFlaskStreamlitGitHub Actions

Additional programming languages

CC++C#JavaHTMLCSSMySQL

From an electrical and electronics engineering background, not from the portfolio projects.

Workintech Data Scientist Certificate

Workintech

Completes the advanced data science and AI phase of the Data Pro Program: supervised learning, feature engineering, unsupervised methods, NLP, deep learning, transformers and LLMs, and reinforcement learning.

Sep 2026

Workintech Data Analyst Certificate

Workintech

Role-specific analytics training completed through the Workintech Data Pro Program.

Jun 2026

Introduction to Large Language Models

Google Cloud Training Online

Backs the transformers, LLM API, LangChain and RAG work in the completed Data Pro curriculum.

Jan 2025

Introduction to Responsible AI

Google Cloud Training Online

Responsible model development and use.

Jan 2025

Digital Transformation with Google Cloud

Google Cloud Training Online

Cloud-enabled business and analytics transformation.

Jan 2025

Fundamentals of Software Architecture for Big Data

University of Colorado Boulder

Data architecture and scalable-system foundations.

Mar 2024

Introduction to Process Mining

IBM

Operational process analysis and improvement context.

Jan 2024

The AI Ladder: A Framework for Deploying AI in your Enterprise

IBM

Enterprise AI adoption, operating-model, and delivery context.

Jan 2024

Education

2021

Eskisehir Osmangazi University

BSc, Electrical and Electronics Engineering

2020

Anadolu University

BBA, Business Administration

1 March - 1 October 2026

Workintech

Data Pro Program, completed | Data Analyst Certificate Jun 2026, Data Scientist Certificate Sep 2026

Training track

The curriculum behind the case studies.

Workintech Data Pro Program, 1 March - 1 October 2026, completed, 24 sprints, completed end to end. The lead case studies do not evidence all of it — recurrent networks, building a transformer, and reinforcement learning appear only here.

June 2026

Data Analyst Certificate

September 2026

Data Scientist Certificate

Exercise repositories on GitHub

Colab challenge notebooks

Data Analyst phase

14 exercises

Python, data handling and SQL

Python and OOPpandasMulti-file joinsEDASQL queryingWeb scrapingPlotly

Olist CEO request Multiple files with pandas Exploratory analysis

Data Analyst phase

8 exercises

Statistics and probability

Linear algebraLaw of large numbersCentral limit theoremMarkov chainsBayes theoremRegression inferenceLogistic regression

CLT in real life Markov chains Logit

Data Scientist phase

14 exercises

Supervised learning fundamentals

Linear and logistic regressionKNNSVMRegularizationLoss functionsGradient descent and solversLearning curvesThreshold tuning

SVM and kernels Gradient descent Learning curves

Data Scientist phase

7 exercises

Feature engineering and unsupervised learning

Custom scikit-learn transformersPipeline tuningPCAKMeansRecommender systemsKaggle competition workflow

Custom transformer PCA face recognition Kaggle competition

Data Scientist phase

12 exercises

NLP, deep learning and sequence models

Text classificationTopic modelling with LDATensorFlow and KerasMulticlass classificationImbalanced dataRNNsTime-series forecasting

First RNN LDA topic modelling Credit-card challenge

Data Scientist phase

6 exercises

Transformers, LLMs and RAG

Transformer architectureTensor operationsPrompting small language modelsLLM APIsLangChainRetrieval-augmented generation

RAG with LangChain First transformer LLM APIs

Data Scientist phase

4 exercises

Reinforcement learning

Q-learningStable-Baselines3DQNPPOGym environmentsCustom environments

Cliff walking Car racing Custom environment

Course exercises, kept as breadth evidence and tiered below the case studies. Each module links a representative sample rather than every repository.

Public proof

The external surfaces that support the portfolio.

These channels act as the proof layer around the site itself: code on GitHub, competition standing on Kaggle, a published model on Hugging Face, professional context on LinkedIn, demos on YouTube, and the current CV for role-oriented review.

Professional context, selected featured work, and career-facing positioning.

Open link

GitHub

Source code, repository structure, and project-level implementation proof.

Open link

Kaggle

Competition and notebook profile. Live competitions: Enveda CASMI 2026 (public score 0.422, rank 353 of 2,797 on 2026-10-08) and Playground S6E10 (public 0.96177, rank 34 of 1,126). It is also the source for the MSCapital standing the case study quotes, including the part that is below the field median. The profile contains competition participation and research notebooks; individual notebooks should only be considered public evidence when their current URLs work. Earlier IEEE-CIS leakage and NEU robustness notebook links are unavailable; repository-based, inspectable evaluation reports are used instead for these projects.

Open link

Hugging Face

Model hub profile. It hosts five models, each with a model card that states where it fails: the APTOS retinopathy grader, the MSCapital ensemble, the AI4I machine-failure model, the Transfermarkt value-change model and the NEU steel-defect ResNet-18. The APTOS, AI4I and NEU models also run as in-browser demos on Hugging Face Spaces.

Open link

YouTube

Video channel. Case-study pages do not link individual videos; their evidence is the live demos, repositories and documents.

Open link

CV

Current one-page Data Scientist CV as a PDF, served from this site.

Open link

Contact

Reach out if there is a clear fit or a thoughtful reason to talk.

I am open to concise conversations around data science, applied machine learning, model evaluation, analytics engineering, project walkthroughs, and role discussions that genuinely match the direction of the work shown here.

Send a message

Use the form or email me directly.