Skip to content

Emre Çancıoğlu

I'm a Digitalization & AI Supervisor.|

About

This section highlights my dedication to driving digital transformation and enhancing operational efficiency.

Emre Çancıoğlu

Digitalization & AI Supervisor | Industrial Automation.

Results-driven Digitalization & AI Supervisor with over 9 years of experience leading digital transformation and applied-AI initiatives on the industrial production floor — currently overseeing OT digitalization and AI strategy across 3 manufacturing facilities.

Birthday: August 10, 1995
City: İzmir/Türkiye
Age: 31
Degree: Master
Freelance: Available

Managing a 2-person engineering team, coordinating across 10+ cross-functional departments, and leading the company-wide AI Ambassador program. Delivers measurable impact through predictive maintenance, computer-vision-based safety systems (contributing to 4 consecutive years of zero workplace accidents in the areas where they were deployed), GenAI-powered automation, and energy & sustainability analytics. Deep hands-on expertise with PLC, SCADA, and industrial protocols (OPC UA, MQTT, Profinet), balanced with strong software/cloud and leadership capability (Node.js, React, Azure/AWS, Kubernetes). Recognized with multiple 1st-place finishes and an Innovation Special Award across İnci Holding's group companies, and the author of 5 scientific publications in applied machine learning and industrial automation.

  • 9+Years Experience
  • 3Facilities
  • 2Person Team
  • 57%Cost Reduction
  • 5Publications
  • 6Awards

Skills

My skills encompass a wide range of expertise in digital transformation, operational optimization, and the implementation of advanced technological solutions.

OT Systems

PLC (Siemens, Allen-Bradley, Mitsubishi)SCADADCSHMIMESOPC UAMQTTModbus (TCP/RTU)ProfinetEtherNet/IPCC-LinkKepwareUnified NamespaceDigital TwinISA-95NATVLAN SegmentationIndustrial FirewallsIEC 62443

AI & Data

Predictive MaintenanceComputer VisionMachine LearningDeep LearningGenAI/LLM Automation (RAG, n8n, Copilot, GPT, Claude)MLOpsEdge ComputingETL/Stream ProcessingSPCGrafanaPower BIData Analytics

Leadership

Team Leadership & MentoringCross-functional CollaborationProject & Budget ManagementBias for Action

DevOps

AzureAWSDockerKubernetesRancherTerraformJenkinsGitHub ActionsNGINXMS IISPM2Git

Development

Node.jsExpress.jsDjangoFastAPIREST APIKafkaReactReduxJavaScriptTypeScriptPythonMS SQLMongoDBPostgreSQLDynamoDBElasticsearch

Languages

Turkish (Native)English (Professional)

Resume

My resume highlights my expertise in delivering impactful solutions that accelerate digital transformation and enhance operational efficiency.

Honors & Awards

Cevdet İnci Incentive Awards are held across İnci Holding's group companies (5 sister companies, 9+ facilities); İnci GS Yuasa Stars is company-specific.

Sustainability Business Awards

2025

İstanbul/TÜRKİYE

  • Finalist (Technology & AI Category)

Cevdet İnci Incentive Awards

2025

İzmir/TÜRKİYE

  • 1st Place

İnci GS Yuasa Stars

2025

Manisa/TÜRKİYE

  • 3rd Place

Cevdet İnci Incentive Awards

2024

İzmir/TÜRKİYE

  • 1st Place
  • Innovation Special Award

İnci GS Yuasa Stars

2024

Manisa/TÜRKİYE

  • 1st Place

Cevdet İnci Incentive Awards

2023

İzmir/TÜRKİYE

  • 1st Place

Presentations & Publications

13th International European Conference on Interdisciplinary Scientific Research

May 2026

Tirana, ALBANIA

Writer | Researcher

Topic: ML regression models (XGBoost, LightGBM, GBM) forecasting electricity and compressed-air consumption on battery production lines, achieving R² up to 0.944 for short-term forecasts.

R&D & Innovation 2024

Dec 2024

Manisa, TÜRKİYE

Writer | Researcher

Topic: Data Integration for Industry 4.0 and IoT: Unified Namespace-Based Digital Transformation with MQTT, OPC UA, and Node.js.

Human-Computer Interaction Optimization and Robotic Applications

Jun 2021

TÜRKİYE

Writer | Researcher

Topic: Fault Detection and Diagnosis in Process Control Systems using Machine Learning (ensemble learning) Methods from Poincaré Plot Measurements, improving classification accuracy to 89.5% on the Tennessee Eastman Process.

International Medical Device Conference

Nov 2020

Antalya, TÜRKİYE

Writer | Researcher

Topic: Heart Sounds Analysis and Classification Based on Long-Short Term Memory: an LSTM model classifying phonocardiogram recordings into Normal, Murmur, Extrasystole, and Artifact categories with 79.0% accuracy (Journal of Intelligent Systems with Applications, 3(1), 2020).

Journal of Intelligent Systems with Applications

2020

İzmir, TÜRKİYE

Writer | Researcher

Topic: Design and Implementation of Digital Filters for ECG Data Based on Field Programmable Gate Array and MATLAB: four digital filters (low-pass, high-pass, band-pass, band-stop) built on FPGA with MATLAB System Generator (3(1), 2020).

Certifications

"Geleceği Sahipleniyoruz" Ciddi Oyunlar™ ile Geleceği Sahiplenen İnovasyon Projeleri

Jul 2026

41 North Business School

"Geleceği Sahipleniyoruz" Blackbox Business Challenge™

Jun 2026

41 North Business School

ISO 9001:2015 & IATF 16949:2016 Training

Feb 2023

KALMER Kalite Yönetim Merkezi

ISO 9001:2015, ISO 14001:2015, ISO 45001:2018

Nov 2021

N-Sistem

Hydraulics & Pneumatics Development and Adaptation Training (64 hours)

Feb 2020

Manisa Yunusemre Halk Eğitim Merkezi

Database Management – SQL Development and Adaptation Training (56 hours)

Dec 2019

Manisa Yunusemre Halk Eğitim Merkezi

Computer-Aided Design (AutoCAD) (160 hours)

Dec 2017

İzmir Halk Eğitim Merkezi

Programmable Logic Control (PLC) (296 hours)

Dec 2017

İzmir Halk Eğitim Merkezi

Professional Experience

Digitalization & AI Supervisor

Jan 2026 - Present

İnci GS Yuasa / İnci Akü, Manisa/TÜRKİYE

  • Leading OT digitalization and applied-AI strategy across 3 production facilities; managing a 2-person engineering team and coordinating across 10+ cross-functional departments. Delivered 15+ process-improvement projects since taking on the role.
  • Scaled the OT data integration backbone to 300+ managed data channels and database connections (up from ~200) across PLC, SCADA, MES, and historian systems — integrating 36 new production lines/machines in 2026 alone (furnaces, mills, assembly, filling, and welding stations), surpassing the year's 20% connectivity target to reach 25.76% facility-wide machine coverage, and building a self-service OPC UA-based alerting platform that lets any user configure real-time custom alarms across all connected machines.
  • Grew the applied-AI portfolio to 4 active predictive-maintenance/quality models and 25 AI-driven automation projects, including RAG-based knowledge systems and GenAI/multi-agent workflows (n8n, Microsoft Copilot, GPT, Claude) that cut day-long reporting cycles down to seconds; leading the AI Ambassador (AI Elçileri) program, driving every department to build and own its own AI agent.
  • Deployed real-time air-quality/particulate monitoring in areas with high lead-dust exposure, contributing to a 58.62% reduction in workers' blood lead levels; also added real-time anomaly detection on production/utility systems that helped prevent a potential fire incident.
  • Integrated two YOLO-based computer-vision safety models: AI-ODS, which auto-stops robots when a person enters their work envelope, and AI-FDS, which uses real-time light-projection warnings to prevent forklift-pedestrian and forklift-forklift collisions at blind spots — sustaining zero workplace accidents for 4 consecutive years in the areas where they were deployed.
  • Retrofitted 200+ electricity, water, and natural-gas meters to remote-reading, built the underlying database, and launched an energy dashboard visualizing machine-level consumption, per-unit-product energy use, and carbon footprint — integrated with SAP and preparing the facility for ISO 50001 certification.
  • Built a compressed-air leak anomaly-detection algorithm fusing compressor pressure data, real-time MES production signals, and SAP machine-level production-confirmation data, enabling early leak intervention and reducing compressor energy consumption.
  • Added ML-based energy forecasting driven by planned production volumes (XGBoost, LightGBM, GBM regression, R² up to 0.944), co-authored as peer-reviewed research at the 13th International European Conference on Interdisciplinary Scientific Research (Tirana, Albania, May 2026).

Senior Operational Technologies (OT) Engineer

Dec 2022 - Dec 2025

İnci GS Yuasa / İnci Akü, Manisa/TÜRKİYE

  • Architected a Unified Namespace (UNS) unifying SAP, MES, and PLC/SCADA/DCS-based shop-floor equipment — over OPC UA, MQTT, Kafka, Profinet, EtherNet/IP, and CC-Link — with energy analyzers and third-party systems, establishing the facility's digital twin; led a 20-person team to give every battery full-lifecycle process and test traceability from raw material to shipment.
  • Led the design and rollout of the OT network architecture; ran 80+ structured cybersecurity and disaster-recovery operations across VLAN segmentation, NAT, and industrial firewalls aligned to the ISA-95/Purdue model and IEC 62443 to safeguard production continuity and data security.
  • Built a 46-channel, 156-database backbone (MS SQL, MongoDB, PostgreSQL) spanning all 3 production facilities on a custom Node.js/React/Redux/TypeScript stack with ETL/stream-processing pipelines, automating entirely manual shop-floor workflows to lift operational efficiency 19% and data reliability 14%, and cutting OEM/AFM customer data-request turnaround by 90%.
  • Delivered end-to-end visibility via Power BI and custom dashboards showing real-time machine data, per-unit test results, and real-time SPC-based anomaly alerts — further strengthening the digital twin and powering the Digital Product Passport; accelerated reporting by 60%, saving $156K annually.

Software Development / Automation Lead

Jan 2022 - Nov 2022

DVD Valves, Manisa/TÜRKİYE

  • Developed a PLC-controlled hydraulic butterfly valve control system with a Modbus TCP-based SCADA platform for international water projects (Italy: 8ר700mm, Dubai: 2ר1400/Ø1500mm, Colombia: 2ר700mm); completed FAT/SAT testing. Brought development in-house instead of outsourcing, cutting costs by 30% ($80k saved).
  • As project lead overseeing the ~₺3.5M TÜBİTAK TEYDEB 1501 budget, led a 3-person core team (coordinating cross-departmental contributors across the project's 9+-person team) building the embedded software and desktop/mobile applications for the pre-charge and ultrasonic metering modules of a smart irrigation hydrant system, cutting unit cost by ~68% versus the outsourced solution and bringing after-sales technical support fully in-house.
  • TÜBİTAK 2209-B industrial consultant (smart agricultural valve project, No. 119B412103138): added LoRa-based remote meter reading.

Software Development / Automation Engineer

Jul 2018 - Dec 2021

DVD Valves, Manisa/TÜRKİYE

  • Designed two autonomous welding robots with PLC/HMI control to deliver a robust, high-quality weld in a sealing-critical zone, cutting welding cycle time by 60% and reducing sealing-related quality defects by 23%.
  • Enabled real-time monitoring via the Focas library for CNC machines and OPC UA for PLC-based equipment, with REST API-based data collection and historical data logging.
  • Modernized outdated, manually-operated control panels and revised PLC software for 10+ production-floor machines (blower test systems, sealing test units with an MS SQL-based desktop application, rubber press, sizing, polishing robot, saw, radial drill, and sandblasting), upgrading their automation level.
  • Enabled real-time, cross-department visibility into production and maintenance operations through MES and ERP integration.

Part-Time Project Engineer

Jul 2017 - Jun 2018

DİMES, İzmir/TÜRKİYE

  • Transitioned a fully manual PET bottle production line to a fully autonomous conveyor system, designing the PLC-based control panel.
  • Developed PLC-SCADA software for mixing tank control, enabling full traceability and digitizing reporting and quality processes, supporting faster detection and root-cause analysis of quality deviations.

Education

MSc. in Electrical & Electronics Engineering

Sep 2019 - Jun 2022

İzmir Katip Çelebi University, İzmir/TÜRKİYE

GPA: 3.64/4.00 (100% English-Medium Program)

Coursework: Statistical Process Control | Data Acquisition and Control | Applied Machine Learning | Artificial Neural Networks.

Thesis: Fault Detection and Diagnosis of the Tennessee Eastman Process Using Statistical Analysis and Machine Learning Methods from Poincaré Plot Measurements.

BSc. in Electrical & Electronics Engineering

Sep 2014 - Jul 2019

İzmir Katip Çelebi University, İzmir/TÜRKİYE

GPA: 3.14/4.00 (100% English-Medium Program)

Coursework: Process Control and Instrumentation | Optimization in Engineering | Industrial Automation | Signals and Systems.

Thesis: Design of a Cylindrical Welding and Polishing Machine with PLC-Based HMI Control.

Services

In my areas of expertise, innovative and comprehensive services are provided to enhance digital transformation and operational efficiency.

OT Digitalization & Unified Namespace

End-to-end OT data integration across PLC, SCADA, MES, historian, and SAP systems over OPC UA, MQTT, Kafka, Profinet, EtherNet/IP, and CC-Link. Unified Namespace architectures and digital twins that give full process and test traceability, backed by secure OT networks aligned to ISA-95/Purdue and IEC 62443.

Applied AI & GenAI Automation

Predictive-maintenance and quality models, RAG-based knowledge systems, and GenAI/multi-agent workflows (n8n, Microsoft Copilot, GPT, Claude) that turn day-long reporting cycles into seconds. Leading the company-wide AI Ambassador program so every department builds and owns its own AI agent.

Computer Vision for Workplace Safety

YOLO-based safety systems such as robot auto-stop when a person enters the work envelope and light-projection warnings against forklift collisions at blind spots — contributing to 4 consecutive years of zero workplace accidents where deployed, alongside real-time air-quality and anomaly monitoring.

Energy & Sustainability Analytics

Remote-reading meter retrofits, machine-level energy dashboards with per-unit consumption and carbon footprint, compressed-air leak detection, and ML-based energy forecasting from planned production volumes — integrated with SAP and supporting ISO 50001 readiness.

Full-Stack Development & Cloud

Custom Node.js/React/Redux/TypeScript platforms with ETL/stream-processing pipelines, REST APIs, and real-time dashboards, deployed on Azure/AWS with Docker, Kubernetes, Terraform, and CI/CD — bridging the shop floor and the cloud.

Recognized Innovation & Research

Multiple 1st-place finishes and an Innovation Special Award at the Cevdet İnci Incentive Awards, İnci GS Yuasa Stars awards, and a Sustainability Business Awards finalist spot (Technology & AI). Author of 5 scientific publications in applied machine learning and industrial automation.